Isabella MORLINI - personale UniMoRe

Nuova ricerca

Isabella MORLINI

Professore Associato
Dipartimento di Comunicazione ed Economia

Pubblicazioni

2024 - Statistica: dalla datificazione dei processi alla previsione [Capitolo/Saggio]
Cavicchioli, Maddalena; Demaria, Fabio; FRANCO VILLORIA, Maria; Frederic, Patrizio; Martini, Maria Cristiana; Montagna, Silvia; Morlini, Isabella
abstract

2023 - Characterization of annual urban air temperature changes with special reference to the city of Modena: a comparison between regression models and a proposal for a new index to evaluate relationships between environmental variables [Articolo su rivista]
Morlini, Isabella; Albertson, Sean; Orlandini, Stefano
abstract

It is commonly recognized that the observed increase in global mean annual air temperature is strongly related to the increase in global carbon dioxide concentration C, and that both these variables are related to global development. It remains, however, unclear the degree to which local mean annual urban air temperature T is affected by local variables such as annual precipitation depth P and urban area extent A. This study assumes that A is a proxy of local development and C is a proxy of global development and investigates the commingled effects of A, P, and C on T by using long-term annual data observed over the years 1881–2019 from the Modena Observatory in Italy. Linear relationships between T, C and A are found to be spurious since all these series have a monotonic increasing trend with time. Parametric analytic models like logistic functions are found to lack flexibility. Smoothing splines can only give insights into the strength of the relationships but not on their shape defining the functional relationship between variables. Advanced nonlinear models like generalized additive models, instead, are found to combine flexibility in a parametric form, and appear therefore to be suitable models for explaining the complex relationships between A, P, and C on T. The different models are evaluated using traditional goodness of fit statistics like R2, AIC, BIC, and a new index of relation IR which is introduced to jointly evaluate the goodness-of-fit of relationships between variables that may either be dependent or independent.

2023 - Modelling local climate change using site-based data [Articolo su rivista]
Morlini, Isabella; Franco-Villoria, Maria; Orlandini, Stefano
abstract

2023 - STATISTICA: LA SCIENZA CHE MODELLA I DATI. Un'introduzione alle diverse tipologie di dati. [Capitolo/Saggio]
Cavicchioli, M.; Demaria, F.; FRANCO VILLORIA, M.; Frederic, P.; Morlini, I.
abstract

2022 - Che cos’è la statistica? Una prima introduzione alla scienza dei dati [Capitolo/Saggio]
Cavicchioli, Maddalena; FRANCO VILLORIA, Maria; Frederic, Patrizio; Morlini, Isabella
abstract

“Il portiere della nazionale di calcio ha parato nell’ultimo anno 6 rigori, portando a 0.75 il suo numero medio di rigori parati a partita”, “la maggioranza dei ragazzi dai 6 ai 13 anni non fa un’adeguata colazione la mattina”, “il programma trasmesso ieri in prima serata sul canale 66 del digitale terrestre ha avuto il 15% di share”, “oggi la probabilità di precipitazioni è del 60%”. Sono frasi che inevitabilmente ogni giorno ci vengono proposte nella nostra vita quotidiana. Ovunque, sempre, ci vengono presentati numeri e termini tecnici che possono sembrare strani e di difficile comprensione. Se non impariamo a trasformare le cifre che ci circondano in informazione utile, interpretata e a nostra volta comunicata in modo appropriato, non riusciremo mai a rispondere a domande importanti per le nostre decisioni e a soddisfare le nostre esigenze conoscitive. E se vi dicessimo che la statistica è proprio la materia che permette di decodificare i dati numerici e non numerici e che vi aiuta a valutare in modo intelligente le informazioni che i dati contengono? E se vi dicessimo che la statistica è fondamentale per poter sfruttare appieno le fonti di informazioni facilmente accessibili in questa era digitale? E se vi dicessimo che la statistica è ormai diventata una parte indispensabile della vita e non conoscerla e utilizzarla significa consegnare inevitabilmente il proprio futuro a qualche altra persona? Forse non avreste una preclusione a priori verso questa materia … E magari sareste anche incuriositi di sapere di cosa si occupa la statistica e quali sono le principali nozioni per iniziare a trasformare i dati in informazione utile! Questa pubblicazione, che raccoglie i contributi di docenti universitari ma che utilizza un linguaggio informale e discorsivo, cercherà di soddisfare queste vostre nuove curiosità. Il libro nasce da un’iniziativa di public engagement promossa dall’Dipartimento di Economia Marco Biagi dell’Università di Modena e Reggio Emilia (Unimore) per condividere anche con gli studenti più giovani la scoperta dei benefici e del fascino della statistica, in tutte le sue declinazioni.

2022 - Data and detailed results- LEARNING ONLINE: REMOTE TEACHING AND UNIVERSITY STUDENTS’ ENGAGEMENT [Banca dati]
Russo, Margherita; Alboni, Fabrizio; Morlini, Isabella; Pavone, Pasquale; Sartori, Laura
abstract

Database and detailed results of the paper LEARNING ONLINE: REMOTE TEACHING AND UNIVERSITY STUDENTS’ ENGAGEMENT, published by Statistica Applicata- Italian Journal of Applied Statistics, 2022

2022 - Evaluation of an Online Course Promoting Health and Wellbeing for University Students and Employees. [Articolo su rivista]
Ricci, F.; Modenese, A.; Gobba, F.; Morlini, I.
abstract

Published studies dealing with health promotion activities, such as the improvement of physical activity and healthy eating, for workers and students prove the effectiveness of these preventive interventions. The consequent benefits include better prevention of cardiovascular risk and an improvement in quality of life. Considering this, an intervention aimed at promoting healthy eating and non-sedentary lifestyles has been implemented within an Italian university: the aim of the present research is to evaluate its effectiveness. The intervention consisted of a targeted asynchronous e-learning two-hour course on healthy eating and non-sedentary lifestyles. The attendants were 2004 university students and employees. We conducted two surveys before and after the training intervention, and, through the responses obtained, we evaluated the effectiveness of the intervention. We applied different statistical methods, including unpaired t-tests and nonparametric tests, principal components and cluster analysis. Our results indicate that the post-training knowledge has been significantly improved, compared to that pre-training (7.3 vs. 8.7, p < 0.001). Moreover, the whole sample showed an improved awareness of the importance of healthy behaviors, and perception of the University as an institution promoting a healthy lifestyle. Through the principal components analysis, we identified a unidimensional latent factor named “health and behaviors”. The cluster analysis highlighted that the sub-group reporting the lowest scores at the survey before the training was the one with the highest improvement after the intervention. To the best of our knowledge, this is the first Italian study testing, before and after a health promotion intervention, the knowledge and the attitudes and behaviors towards healthy lifestyles of a group of students and workers. Moreover, we also evaluated the pre- and post-intervention perceived health status, as well as the level of engagement of the attendants, with respect to their colleagues and management in an educational institution promoting wellbeing. The conclusions of our study support the need for further adoption of health promotion training interventions, similar to the one we performed, in order to improve healthy eating and non-sedentary behaviors among workers and students.

2022 - Impact of Emergency Online Classes on Students’ Motivation and Engagement in University During the Covid-19 Pandemic: A Study Case [Relazione in Atti di Convegno]
Morlini, I.
abstract

The Covid-19 pandemic has had dramatic impact on many dimensions of living and studying conditions of students at University. This paper analyses student satisfaction and motivation during the lockdown period and try to understand whether different socio-economic and environmental conditions have influenced needs and demands of students during the emergency online didactics. Drawing from the results of a questionnaire administered to students enrolled in the University of Modena and Reggio Emilia, this research is aimed at describing which factors, beyond the quality and the professionalism of the lecturers and the quality of the education received, influence the satisfaction with the online learning experience and impact on students’ motivations and perceived engagement. Moreover, the study investigates the pandemic’s direct effects on gender differences and inequalities, analysing the obstacles affecting the self- organization of study at home.

2022 - Learning online: Remote teaching and university students’ engagement [Articolo su rivista]
Russo, Margherita; Alboni, Farizio; Morlini, Isabella; Pavone, Pasquale; Sartori, Laura
abstract

The COVID-19 pandemic has had a dramatic impact on many dimensions of living and working conditions, and uncertainties about the developments that we shall still face in the near future. This paper analyses the implications of a forced overnight push to online teaching. Drawing upon an online survey conducted during the 2020 lockdown by the University of Modena and Reggio Emilia, this article describes students’ living and studying conditions revealed by a large set of open and closed questions. The survey provides significant information on the students’ real off-campus conditions, crucial data for the multidimensional analysis by combining non-parametric multivariate analysis of closed questions with textual analyses. It offers important indications about the most useful tools for inclusive teaching across thematic areas and highlights the main difficulties that emerged during the lockdown. Reflections on advantages and disadvantages, strengths and weaknesses in the innovative learning environment set up overnight are offered at a policy level.

2021 - Effects of remote teaching on students’ motivation and engagement: the case of the University of Modena and Reggio Emilia [Relazione in Atti di Convegno]
Morlini, Isabella; Sartori, Laura
abstract

The Covid-19 pandemic has had dramatic impact on many dimensions of living and studying conditions of students at University. This paper analyses student satisfaction and motivation during the lockdown period and try to understand whether different socio-economic and environmental conditions have influenced the needs and the demands of the students for the online didactics. Drawing from the results of a questionnaire administered to students enrolled in the University of Modena and Reggio Emilia, this research is aimed at describing which factors, beyond the quality and the professionalism of the lecturers and the quality of the education received, influence the satisfaction with the online learning experience and impact on students’ motivations and perceived engagement.

2020 - Covid-19 and UNIMORE students: how the emergency changes the study and the university experience [Working paper]
Russo, M.; Alboni, F.; Colombini, S.; Morlini, I.; Pavone, P.; Sartori, L.
abstract

The research report presents the results of the online survey conducted by the University of Modena and Reggio Emilia on the living and study conditions of students in the period 8 April - 2 May 2020. The online survey was answered by 20% of students: an adequate participation rate to capture some significant differences between courses of study. The aim of the survey is to overcome the Covid-19 emergency with greater awareness, analysing the living and study conditions of the students and looking for the most useful tools for inclusive teaching and study that contribute to giving each student the possibility to proceed in their own study and life path in the best possible way. The questionnaire, the survey tools and the results that emerge from the multivariate analysis are presented, both for the closed-answer questions and for the openended questions, on which textual analysis techniques have been applied. The different tools and methods are presented to emphasize the specificity of the dataset, which presents a different level of variability of the answers in the different thematic areas. The report concludes with a series of considerations that we offer to the Unimore community to support the discussion on what we have learned and continue to learn thanks to the voice of the students. It is now clear that the exit from the emergency will be a gradual process. The students' point of view, which we gather from this survey, will be really valuable to plan next year's educational activities and services for students, in order to offer a high quality teaching that responds to different needs and study contexts. Further developments of the analysis may concern other in-depth studies on specific teaching methods (distance and in-presence) and on the evaluation of the living and working conditions of teaching and technical-administrative staff. The basic idea is to read, through the Covid-19 emergency lens, the essential dimensions to improve the quality of teaching highlighted by the empirical survey.

2020 - Covid-19 e studenti UNIMORE: come l’emergenza cambia lo studio e l’esperienza universitaria [Working paper]
Russo, M.; Alboni, F.; Colombini, S.; Morlini, I.; Pavone, P.; Sartori, L.
abstract

2019 - New Fuzzy Composite Indicators for Dyslexia [Capitolo/Saggio]
Morlini, Isabella; Scorza, Maristella
abstract

In this paper, we suggest a method based on fuzzy set theory for te construction of fuzzy synthetic indexes of dyslexia. A few criteria for assigning values to the membership functions are discussed, as well as criteria for defining the weights of the manifest variables.

2018 - Fuzzy method s for the analysis of psychometric data:an application for measuring reading disability [Articolo su rivista]
Morlini, Isabella
abstract

Psychometrics should ideally measure multidimensional concepts like skills, knowledge, abilities, attitudes, personality traits and educational achievement, which cannot be captured by a single variable. In this paper, we suggest a method based on the fuzzy set theory for the construction of a fuzzy synthetic index of the latent psychometric phenomenon, using the set of variables obtained with instruments such as questionnaires or tests. Criteria for assigning values to the membership function as well as criteria for defining the weights of the variables are discussed. For discrete variables, we use a fuzzy quantification method based on the sampling cumulative function. An application regarding the measurement of reading disability in students attending elementary and middle school in Italy is presented.

2017 - Fuzzy Methods for the Analysis of Psychometric Data [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Psychometrics should ideally measure multidimensional concepts like skills, knowledge, abilities, attitudes, personality traits and educational achievement, which cannot be captured by a single variable. In this paper, we suggest a method based on fuzzy set theory for the construction of a fuzzy synthetic index of the latent psychometric phenomenon, using the set of variables obtained with instruments such as questionnaires or tests. Criteria for assigning values to the membership function as well as criteria for defining the weights of the variables are discussed. For discrete variables, we use a fuzzy quantification method based on the sampling cumulative function. An application regarding the measurement of reading disability in students attending elementary and middle school in Italy is presented.

2017 - New fuzzy composite indicators for dyslexia [Relazione in Atti di Convegno]
Morlini, Isabella; Scorza, Maristella
abstract

Composite indicators should ideally identify multidimensional concepts that cannot be captured by a single variable. In this paper, we suggest a method based on fuzzy set theory for the construction of fuzzy synthetic indexes of dyslexia, using the set of manifest variables measured by means of reading tests. A few criteria for assigning values to the membership function are discussed, as well as criteria for defining the weights of the variables. An application regarding the diagnosis of dyslexia in primary and middle school in Italy is presented. In this application, the fuzzy approach is compared with the crisp approach actually used in Italy for detecting dyslexic children in compulsory school.

2017 - Studio dell’incidenza della dislessia nelle scuole elementari e medie in Emilia Romagna e Lombardia [Articolo su rivista]
Morlini, Isabella; Scorza, Maristella
abstract

Lo studio presenta i risultati ottenuti utilizzando un nuovo indicatore composito fuzzy per la disgnosi della dislessia nelle scuole elementari e medie italiane. I dati raccolti mostrano l’utilità dell’utilizzo del nuovo indicatore composito. L’indice dà la possibilità di utilizzare congiuntamente tutti gli aspetti legati all’abilità di lettura (sia la velocita, misurata in tempo di lettura, sia l’accuratezza, misurata in numero di errori). Inoltre, la natura fuzzy dell’indice permette di avere una classificazione sfocata che considera i diversi livelli di gravità della dislessia ed anche i diversi livelli di abilità nella lettura e non segue la suddivisione rigida fra “patologia presente” e “patologia assente”.

2016 - The evolution of the reading profile in children with developmental dyslexia in a regular ortographies [Articolo su rivista]
Zonno, Maria Giuseppa Pia; Scorza, Maristella; Morlini, Isabella; Stella, Giacomo
abstract

Several researchers have demonstrated that dyslexia develops differently in shallow orthographies in terms of accuracy and speed. In fact, slow reading speed persists and accuracy improves. The aim of this study is to investigate the evolution of the specific reading disorder over the years of compulsory education, from primary to upper secondary school. Furthermore, it has the aim to verify if there are different evolutionary trajectories of reading skills in relation to the severity of the disorder. The study was carried out on 71 Italian dyslexic children, according to the diagnostic criteria established by the diagnostic manual ICD – 10 and the Consensus Conference. Two groups were selected: children who met criteria for mild dyslexia (mild dyslexics, with n=36) and a comparison group of moderate-severe dyslexics (n=35). All participants were tested at least twice in two different school grades. Comparisons were made on the average performances in each school grade. The results reveal similar patterns of growth over time in reading ability, with the mild dyslexics group outperforming the moderate-severe dyslexics group. The performance trajectory for the moderate-severe dyslexics shows some plateaus and a decrease in performances in the last year analyzed (1st upper secondary school) while the trajectory for the mild dyslexics always show increases in performances. All subjects show a steady increase in word and text reading speed and a slower improvement in pseudo-word decoding.

2015 - Advances in Statistical Models for Data Analysis [Curatela]
Morlini, Isabella; Minerva, Tommaso; Vichi, Maurizio
abstract

This volume contains peer-reviewed selected contributions presented at the 9th biannual meeting of the Classification and Data Analysis Group (CLADAG) of the Italian Statistical Society that took place in Modena from September 18 to September 20, 2013. The conference brought together not only theoretical and applied statisticians working in Italy but also a number of specialists coming from nine different countries and was attended by more than 180 participants, including those who participated in a special session for young researchers. The conference encompassed 122 presentations organised into two plenary talks, two semi-plenary talks, 11 specialized sessions, 11 contributes sessions, eight coordinate sessions and a poster session. The main emphasis on the selection of the plenary and semi-plenary talks and on the call of papers was put on classification, data analysis and multivariate statistics, to fit the mission of CLADAG. However, many chosen contributions regarded related areas like machine learning, Markov models, structural equation models, statistical modelling in economics and finance, education and social sciences and environment.

2015 - Advances in statistical models for data analysis [Relazione in Atti di Convegno]
Morlini, I.; Minerva, T.; Vichi, M.
abstract

2015 - Assessing decoding ability: the role of speed and accuracy and a new composite indicator to measure decoding skill in elementary grades [Articolo su rivista]
Morlini, Isabella; Stella, Giacomo; Scorza, Maristella
abstract

Tools for assessing decoding skill in students attending elementary grades are of fundamental importance for guaranteeing an early identification of reading disabled students and reducing both the primary negative effects (on learning) and the secondary negative effects (on the development of the personality) of this disability. This article presents results obtained by administering existing standardized tests of reading and a new screening procedure to about 1,500 students in the elementary grades in Italy. It is found that variables measuring speed and accuracy in all administered reading tests are not Gaussian, and therefore the threshold values used for classifying a student as a normal decoder or as an impaired decoder must be estimated on the basis of the empirical distribution of these variables rather than by using the percentiles of the normal distribution. It is also found that the decoding speed and the decoding accuracy can be measured in either a 1-minute procedure or in much longer standardized tests. The screening procedure and the tests administered are found to be equivalent insofar as they carry the same information. Finally, it is found that speed and accuracy act as complementary effects in the measurement of decoding ability. On the basis of this last finding, the study introduces a new composite indicator aimed at determining the student's performance, which combines speed and accuracy in the measurement of decoding ability.

2015 - Cluster Analysis of Three-Way Atmospheric Data [Relazione in Atti di Convegno]
Morlini, Isabella; Orlandini, Stefano
abstract

Classification of meteorological time series is important for the analysis of the climate variability and climate change. The clustering of several years in groups that are homogeneous with reference to the amount of precipitation and to the atmospheric condition, can aid in understanding the structure of precipitation and may be important in developing hydrological models. In this paper we propose a cluster analysis of multivariate time series based on a dissimilarity measure that considers the functional form of the data. The unit to be classified are 148 years, from 1861 to 2008, and the variables are the values of precipitation, the minimum temperature and the maximum temperature in different occasions (days or months) in the province of Modena (Northern Italy)

2015 - Decoding automaticity in reading process and practice. How much influence does summer vacation have on children’s reading abilities in primary school? [Articolo su rivista]
Scorza, Maristella; Boni, Claudia Daria; Zanzurino, Giuseppe G. F.; Scortichini, Francesca; Morlini, Isabella; Stella, Giacomo
abstract

This research intends to investigate the impact of reading practice on children’s read-aloud abilities during the learning phase. In order to assess the importance of reading practice, the researchers have examined the possible adverse consequences arising from the substantial reduction in exercise during the summer vacation. According to the model adopted, groups of children from grade first to fifth in primary school have been given three different standardized tests (lists of words, pseudo-words and a text), in three distinct times of the year (end of school, beginning of school and two months after that). The available literature on the subject demonstrates that summer vacation can have a detrimental impact on maths computation and orthography whereas the results relating to reading abilities seem to be considerably disparate. The outcomes of this research prove that speed and accuracy parameters are affected differently by both the decrease and the increase in reading practice. All assessed classes have shown a regular increase in reading speed, and the suspension of the learning practice does not seem to have influenced the performance significantly. This improvement in reading speed apparently comes with an increase in the percentage of mistakes made after summer vacation, especially in the first classes. Therefore, the comparison of the provided results might suggest the existence of independent mechanisms lying behind the development and automaticity of the two examined factors.

2015 - Letter chain e word chain. Un nuovo strumento di screening per l’identificazione dei bambini con difficoltà di lettura [Articolo su rivista]
Scorza, Maristella; Boni, Claudia Daria; Scortichini, Francesca; Morlini, Isabella; Stella, Giacomo
abstract

Lo strumento ideato da Jacobson (1995), con l’obiettivo di identificare precocemente le difficoltà di lettura all’interno della popolazione svedese, è stato adattato per la lingua italiana. Si tratta di due prove a somministrazione collettiva: una di lettura di lettere (letter chain) e l’altra di lettura di parole (word chain). Al fine di verificare la validità e l’attendibilità del test, sono state proposte a tutto il campione le prove di lettura di parole e non parole tratte dalla DDE-2 (Sartori et al., 1995; 2007) e la lettura di un brano tratto dalla batteria MT (Cornoldi e Colpo, 1995; 1998). I risultati mostrano come, nella prima fase di apprendimento della lettura, i bambini (in particolare quelli di classe prima) individuino più facilmente le lettere (avvalendosi di una strategia di tipo visivo), per poi passare nelle fasi più avanzate della scolarizzazione, e cioè a partire dalla classe terza quando ormai il processo di acquisizione della lettura risulta automatizzato nella maggior parte dei bambini, a una strategia più di tipo lessicale.

2015 - Preface [Prefazione o Postfazione]
Morlini, I.; Minerva, T.; Vichi, M.
abstract

2015 - Preface [Studies in Classification, Data Analysis, and Knowledge Organization] [Relazione in Atti di Convegno]
Morlini, I.; Minerva, T.; Vichi, M.
abstract

2014 - A new procedure to measure children's reading speed and accuracy in Italian [Articolo su rivista]
Morlini, Isabella; Stella, Giacomo; Scorza, Maristella
abstract

Impaired readers in primary school should be early recognized, in order to asses a targeted intervention within the school and to start a teaching that respects the difficulties in learning to read, to write and to perform calculations. Screening procedures inside the primary schools aimed at detecting children with difficulties in reading, are of fundamental importance for guaranteeing an early identification of dyslexic children and reducing both the primary negative effects - on learning - and the secondary negative effects - on the development of the personality - of this disturbance. In this study we propose a new screening procedure measuring reading speed and accuracy. This procedure is very fast (it is exactly one minute long), simple, cheap and can be provided by teachers without technical knowledge. On the contrary, most of the currently used diagnostic tests, are about 10 minutes long and must be provided by experts. These two major flaws prevent the widespread use of these tests. On the basis of the results obtained in a survey on about 1500 students attending primary school in Italy, we investigate the relationships between variables used in the screening procedure and variables measuring speed and accuracy in the currently used diagnostic tests in Italy. Then, we analyze the validity of the screening procedure from a statistical point of view and with an explorative factor analysis we show that reading speed and accuracy seem to be two separate symptoms of the dyslexia phenomenon.

2014 - Some experimental results on the role of speed and accuracy of reading in psychometric tests. [Capitolo/Saggio]
Morlini, Isabella; Stella, Giacomo; Scorza, Maristella
abstract

According to the Italian Parliament act (n. 170/2010) that recognizesdyslexia as a physical disturbance, of neurobiological origin, dyslexic children in primary school should be early recognized, in order to asses a targeted intervention within the school and to start a teaching that respects the difficulties in learning to read, to write and to perform calculations. Screening procedures inside the primaryschools aimed at detecting children with difficulties in reading, are not so common in Italy as in other European countries. Nevertheless, screening procedures are of fundamental importance for guaranteeing an early detection of dyslexic children and reducing both the primary negative effects - on learning - and the secondary negative effects - on the development of the personality - of this disturbance. In thisstudy we analyze the validity, from a statistical point of view, of a screening procedure recently proposed in the psychometric literature (Stella et al., 2011). This procedure is very fast (it is exactly one minute long), simple, cheap and can be dispensed by teachers without psychometric experience. On the contrary, the currentlyused tests are much longer and must be provided by skilled teachers. These two major flaw prevent the widespread use of these tests. If the new procedure is found to be reliable, it can be provided to each student in primary school and it can also be repeated in time, in order to monitor the children difficulties. The validity of the procedure and the benchmark with two currently used tests are studied on the thebasis of the results of a survey on about 1500 students attending primary school.

2013 - Cladag 2013. 9th Meeting of the Classification and Data Analysis Group. Book of Abstracts [Curatela]
Minerva, Tommaso; Morlini, Isabella; F., Palumbo
abstract

The 9th biennal international meeting of the Classification and Data Analysis Group of the Italian Statistical Society took place at the University of Modena and Reggio Emilia, from September 18th to September 20th, 2013. The present book contains the short papers presented during this meeting.

2013 - Cluster analysis of three-way atmospheric data [Abstract in Atti di Convegno]
Morlini, Isabella; Orlandini, Stefano
abstract

2013 - Fuzzy composite indicators: an application for measuring customer satisfaction [Capitolo/Saggio]
S., Zani; M. A., Milioli; Morlini, Isabella
abstract

Composite indicators should ideally measuremultidimensional concepts which cannot be captured by a singlevariable. In this paper, we suggest a method based on fuzzy settheory for the construction of a fuzzy synthetic index of a latentphenomenon (e.g. well-being, quality of life, etc.), using a setof manifest variables measured on different scales (quantitative,ordinal and binary). A few criteria for assigning values to themembership function are discussed, as well as criteria fordefining the weights of the variables. For ordinal variables, wepropose a fuzzy quantification method based on the samplingcumulative function and a weighting system which takes intoaccount the relative frequency of each category. An applicationregarding the results of a survey on the users of a contact centeris presented.

2013 - Pappagallo Lalla [Software]
Stella, Giacomo; Burani, Cristina; Gallo, Daniela; Morlini, Isabella
abstract

CD per l'identificazione precoce di difficoltà fonologiche e laboratori per lo sviluppo linguistico da 3 a 5 anni

2013 - Simulation experiments for similarity indexes between two hierarchical clusterings [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Morlini and Zani (2012) have proposed a new dissimilarity indexfor comparing two hierarchical clusterings on the basis of thewhole dendrograms. They have presented and discussed its basicproperties and have shown that the index can be decomposed intocontributions pertaining to each stage of the hierarchies. Then,they have obtained a similarity index $S$ as the complement to oneof the suggested distance and have shown that its singlecomponents $S_k$ obtained at each stage $k$ of the hierarchies canbe related to the measure $B_k$ suggested by Fowlkes \& Mallows(1983) and to the Rand index $R_k$. In this paper, we reportresults of a series of simulation experiments aimed at comparingthe behavior of these new indexes with other well-establishedsimilarity measures, over different experimental conditions. Thefirst set of simulations is aimed at determining the behavior ofthe indexes when the clusterings being compared are unrelated. Thesecond set tries to investigate the robustness to different levelsof noise

2013 - Using latent variables in model based clustering: an e-government application [Capitolo/Saggio]
Morlini, Isabella
abstract

2013 - Variable selection in cluster analysis: an approach based on a new index [Relazione in Atti di Convegno]
Morlini, Isabella; Zani, S.
abstract

In cluster analysis, the inclusion of unnecessaryvariables may mask the true group structure. For the selection ofthe best subset of variables, we suggest the use of two overallindices. The first index is a distance between two hierarchicalclusterings and the second one is a similarity index obtained asthe complement to one of the previous distance. Both criteria canbe used for measuring the similarity between clusterings obtainedwith different subsets of variables. An application with a realdata set regarding the economic welfare of the Italian Regionsshows the benefits gained with the suggested procedure.

2012 - A latent variables approach for clustering mixed binary and continuous variables within a Gaussian mixture model [Articolo su rivista]
Morlini, Isabella
abstract

For clustering objects, we often collect not only continuous variables, but binary attributes as well. This paperproposes a model-based clustering approach with mixed binary and continuous variables where each binaryattribute is generated by a latent continuous variable that is dichotomized with a suitable threshold value, andwhere the scores of the latent variables are estimated from the binary data. In economics, such variables arecalled utility functions and the assumption is that the binary attributes (the presence or the absence of a publicservice or utility) are determined by low and high values of these functions. In genetics, the latent responseis interpreted as the ‘liability’ to develop a qualitative trait or phenotype. The estimated scores of the latentvariables, together with the observed continuous ones, allow to use a multivariate Gaussian mixture modelfor clustering, instead of using a mixture of discrete and continuous distributions. After describing the method,this paper presents the results of both simulated and real-case data and compares the performances of themultivariate Gaussian mixture model and of a mixture of joint multivariate and multinomial distributions.Results show that the former model outperforms the mixture model for variables with different scales, bothin terms of classification error rate and reproduction of the clusters means.

2012 - A new class of weighted similarity indices using polytomous variables [Articolo su rivista]
Morlini, Isabella; S., Zani
abstract

We introduce new similarity measures between two subjects, withreference to variables with multiple categories. In contrast totraditionally used similarity indices, they also consider thefrequency of the categories of each attribute in the sample. Thisfeature is useful when dealing with rare categories, since itmakes sense to differently evaluate the pairwise presence of arare category from the pairwise presence of a widespread one. Aweighting criterion for each category derived from Shannon'sinformation theory is suggested. There are two versions of theweighted index: one for independent categorical variables and onefor dependent variables. The suitability of the proposed indicesis shown in this paper using both simulated and real world datasets

2012 - An overall index for comparing hierarchical clusterings [Relazione in Atti di Convegno]
Morlini, Isabella; S., Zani
abstract

In this paper we suggest a new index for measuring thedistance between two hierarchical clusterings. This index can bedecomposed into the contributions pertaining to each stage of thehierarchies. We show the relations of such components with thecurrently used criteria for comparing two partitions. We obtain asimilarity index as the complement to one of the suggesteddistance and we propose its adjustment for agreement due tochance. We consider the extension of the proposed distance andsimilarity measures to more than two dendrograms and their use forthe consensus of classification and variable selection in clusteranalysis.

2012 - Dissimilarity and similarity measures for comparing dendrograms and their applications [Articolo su rivista]
Morlini, Isabella; S., Zani
abstract

In this paper we propose a new index Z for measuring the dissimilaritybetween two hierarchical clusterings (or dendrograms). This index is a metric since it satisfies the axioms of non-negativity, symmetry and triangle inequality. A desirable property of this index is that it can be decomposed into the contributions pertaining to each stage of the hierarchies. We show the relations of such components with the currently used criteria for comparing two partitions.We obtain a global similarity index as the complement to one of the suggested dissimilarity and we derive its adjustment for agreement due to chance. We obtain similarity indexes pertaining to each stage of the hierarchies as the complement to one of the additive parts of the global distance Z. We consider the use of the proposed distance for more than two dendrograms and its use for the consensus of classifications and variable selection in cluster analysis. A series of simulation experiments and an application to a real data set are presented.

2012 - Fuzzy methods and satisfaction index [Capitolo/Saggio]
S., Zani; M. A., Milioli; Morlini, Isabella
abstract

This chapter develops a framework that uses fuzzy set theory in order to measure customersatisfaction, starting from a survey with several questions. The basic concepts of the theoryof the fuzzy numbers are briefly described. A criterion based on the sampling cumulativefunction, which assigns values to the membership function with reference to each quantitative,ordinal and binary variable, is suggested.Weighting and aggregation operators for the variablesare considered. An application to ABC survey data shows the usefulness of the fuzzy setapproach: the gradual transition from very dissatisfied to really satisfied customers is capturedby fuzzy composite indices. The comparison with the classical methods for the measurementof customer satisfaction highlights the advantages of the suggested criterion from both thetheoretical and operational points of view.

2012 - Il TRPS: nuovi indici psicometrici e predittività dello strumento per lo screening precoce di lettura [Articolo su rivista]
G., Zanzurino; Stella, Giacomo; Morlini, Isabella; Scorza, Maristella; Scortichini, Francesca
abstract

All'ingresso della scuola primaria i bambini presentano un'elevata eterogeneità nello sviluppo, dovuta a differenti livelli di maturazione cognitiva e linguistica. Tale variabilità rende molto difficile differenziare un disturbo di apprendimento da un semplice ritardo di maturazion. Nella ricerca vengono presentati una nuova standardizzazione e nuovi indici psicometrici del test TRPS per il rilevamento di difficoltà di lettura nelle classi prima e seconda della scuola primaria. I risultati, pur confermando alcuni limiti dello strumento TRPS, ne rilevano le potenzialità predittive, soprattutto nelle prime fasi d'acquisizione e nelle condizioni particolarmente severe del disturbo di lettura.

2012 - La diagnosi di Dislessia e Disortografia Evolutiva nei bambini bilingui (L2). Evidenze sul ruolo del lessico [Articolo su rivista]
F. G., Giuseppe Zanzurino; Francesca, Scortichini; Stella, Giacomo; Morlini, Isabella; Scorza, Maristella
abstract

La ricerca presentata pone dei quesiti in merito agli strumenti e ai criteri diagnostici utilizzati per la diagnosi di DSA nei bambini bilingui, riflettendo sulle modalità più utili per discernere le questioni educative (ad esempio, presenza di bambini con una limitata conoscenza della lingua italiana) da quelle di effettivo disturbo, per non correre il rischio di applicare etichette neuropatologiche a situazioni che, invece, patologiche non sono. I risultati della ricerca evidenziano che i bambini bilingui possono essere penalizzati se vengono valutati attraverso prove di tipo lessicale tarate su un campione italiano ma, nello stesso tempo, sembrano offrire anche una via alternativa per far diminuire il rischio di diagnosticare falsi positivi.

2012 - La diagnosi di dislessia e disortografia evolutiva nei bambini bilingui (L2) [Articolo su rivista]
F., Scortichini; Stella, Giacomo; Morlini, Isabella; G., Zanzurino; Scorza, Maristella
abstract

La ricerca presentata pone dei quesiti in merito agli strumenti e ai criteri diagnostici utilizzati per la diagnosi di DSA nei bambini bilingui, riflettendo sulle modalità più utili per discernere le questioni educative da quelle di effettivo disturbo, per non correre il rischio di applicare etichette neuropatologiche a situazioni che, invece, patologiche non sono. I risultati della ricerca evidenziano che i bambini bilingui possono essere penalizzati se vengono valutati attraverso prove di tipo lessicale tarate su un campione italiano.

2012 - SPILLO: un nuovo strumento per l’identificazione della lentezza nella lettura orale [Articolo su rivista]
Scorza, Maristella; Stella, Giacomo; Morlini, Isabella
abstract

The act of Parliament n. 170 (approved the 8th October 2010), on «the newstatutory law for learning disorders affecting the scholastic population» states thatdyslexia is a neurobiologically based dysfunction, which makes learning to read, to write and to perform calculations for intelligent children who do not have any othertypes of disorder, very difficult. According to this act, those who teach a dyslexicchild in school should respect the pace and the learning styles of the individual andshould include a system of assessment that takes into account the different skillareas of student. The early detection of children in primary school becomes of fundamentalimportance. Procedures for recognizing dyslexic children are based on readingperformance. In this paper a new screening is presented. Its new features are thetime saving (it is exactly one minute long), cheap and very easy to implement.If this screening procedure is found to be reliable, it can be provided to allstudents by the teachers, since it does not require skilled people. The new test wasadministred to 1500 primary school students to obtain standard measures. Reliabilityof the new screening task was tested matching results with performances of thesame subjects on two standardized reading test. This study allows us to define thethreshold values in the variables used in the new screening procedure and to discussthevalidity of the threshold values currently used in the variables of the benchmarktests.

2011 - Fuzzy methods and satisfaction indices [Capitolo/Saggio]
Sergio Zani, M. A. M.; Morlini, I.
abstract

This chapter develops a framework that uses fuzzy set theory in order to measure customer satisfaction, starting from a survey with several questions. The basic concepts of the theory of the fuzzy numbers are briefly described. A criterion based on the sampling cumulative function, which assigns values to the membership function with reference to each quantitative, ordinal and binary variable, is suggested. Weighting and aggregation operators for the variables are considered. An application to ABC 2010 annual customer satisfaction survey data shows the usefulness of the fuzzy set approach: the gradual transition from very dissatisfied to really satisfied customers is captured by fuzzy composite indices. The comparison with the classical methods for the measurement of customer satisfaction highlights the advantages of the suggested criterion from both the theoretical and operational points of view.

2011 - Mixed mode data clustering: an approach based on tectrachoric correlations [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

In this paper we face the problem of clustering mixedmode data by assuming that the observed binary variables aregenerated from latent continuous variables. We perform a principalcomponents analysis on the matrix of tetrachoric correlations and wethen estimate the scores of each latent variable and construct adata matrix with continuous variables to be used in fully Guassianmixture models or in the k-means cluster analysis. The calculationof the expected a posteriori (EAP) estimates may proceed by simplyconsidering a limited number of quadrature points. Results on asimulation study and on a real data set are reported.

2011 - New weighed similarity indexes for market segmentation using categorical variables [Relazione in Atti di Convegno]
Morlini, Isabella; S., Zani
abstract

In this paper we introduce new similarity indexes forcategorical data with nominal scale. In contrast to traditionallyused similarity measures, they also consider the frequency of themodalities of each attribute in the sample. This feature is usefulwhen dealing with rare categories, since it makes sense todifferently evaluate the pairwise presence of a rare category fromthe pairwise presence of a widespread one. We also propose aspecific weighted index for dependent categorical variables. Thesuitability of the proposed measures from a marketing researchperspective is shown using two real world data sets.

2011 - SPILLO Strumento per l'identificazione della lentezza nella lettura orale [Software]
Stella, Giacomo; Scorza, Maristella; Morlini, Isabella
abstract

SPILLO è un software che consente di verificare in 1 minuto le abilità di lettura dei bambini della scuola primaria (dalla prima alla quinta) tramite una procedura informatizzata che restituiisce immediatamente i risultati

2011 - Simulation experiments for an overall similarity index between two hierarchical clusterings [Abstract in Atti di Convegno]
Morlini, Isabella; S., Zani
abstract

In Morlini and Zani (2011) we have proposed a new dissimilarity index for comparingtwo hierarchical clusterings, on the basis of the whole dendrograms. We havepresented and discussed its basic properties and we have shown that the index canbe decomposed into contributions pertaining to each stage of the hierarchies. Wehave shown the relation of each component of the index with the currently usedcriteria for comparing two partitions, namely the Rand index and the simple matchingcoefficient. We have obtained a similarity index S as the complement to oneof a suggested distance measure and we have shown that its single components Skobtained at each stage k of the hierarchies can be related to the measure Bk suggestedby Fowlkes & Mallows (1983). Finally, we have dealt with the adjustment ofthe similarity index Sk for agreement due to chance. In this paper we report resultsof a series of Monte Carlo experiments aimed at comparing the behavior of S, Skand other similarity measures over different experimental conditions. The first setof simulations is aimed at determining the behavior of the indexes when the clusteringsbeing compared are unrelated. The second set of simulations tries to investigatethe robustness of the indexes with respect to different level of noise.

2011 - Training lessicale nella Dislessia e Disortografia Evolutiva : analisi dell’efficacia di 2 nuovi trattamenti per il potenziamento della letto-scrittura [Articolo su rivista]
F., Scortichini; Stella, Giacomo; Morlini, Isabella
abstract

L’articolo presenta l’analisi di efficacia di due nuovi trattamenti di tipo lessicale per la Disortografia e la Dislessia Evolutiva. I programmi di potenziamento proposti, sono costituiti da un insieme di brani estratti da libri per la scuola primaria di primo grado ed ordinati rispettando principalmente la loro complessità sintattico-grammaticale (Indice di Gulpease).Lo studio ha coinvolto 53 soggetti frequentanti dal terzo anno della scuola primaria al secondo anno della scuola secondaria di primo grado con diagnosi di Disturbo Specifico dell’Apprendimento (dislessia e/o disortografia). La ricerca mostra un significativo miglioramento delle abilità sottoposte a training (lettura e/o scrittura) alle prove di valutazione utilizzate: lettura e/o scrittura di parole, non parole e testo.

2010 - A non linear and non parametric approach for ground level air pollutants forecasting [Articolo su rivista]
Morlini, Isabella
abstract

One of the main concerns in air quality management is to forecast pollutant concentrationboth to satisfy needs of public information, to predict air quality indexes and to prevent excessivepollutants concentration having negative effects on vegetation and human health. In someregions, forecasting high concentrations lead to public warning and emergency traffic restrictionsaimed at reducing pollution emission due to car fuel. In this work we describe a non parametricand nonlinear predictive model for ground level pollutants concentration. This model developsapproximated confidence intervals with heteroskedastic conditional variance taking as inputspast values of the pollutants and, eventually, covariates such as meteorological factors orpollutant precursors. In this work we present the model and results on a real data setaapplication taking the point of view that most of the necessary information for prediction iscontained in the series itself. However, the theory may be extended straightforward to inputvariables like rain and temperature. In the application we consider several series of daily valuesof ground level air pollutants and we perform short term forecasting. The model may be also usedfor long term forecasting by considering, for example, the series of weekly or monthly averages.

2010 - Comparing approaches for clustering mixed mode data:an application in marketing research [Relazione in Atti di Convegno]
Morlini, Isabella; S., Zani
abstract

Practical applications in marketing research often involve mixtures of categorical and continuous variables. For the purpose of clustering, a variety of algorithms has been proposed to deal with mixed mode data. In this paper we apply some of these techniques on two data sets regarding marketing problems. We also propose an approach based on the consensus between partitions obtained by considering separately each variable or subsets of variables having the same scale. This approach may be applied to data with many categorical variables and does not impose restrictive assumptions on the variable distribution. We finally suggest a summarizing fuzzy partition with membership degrees obtained as a function of the classes determined by the different methods applied.

2010 - Fuzzy Composite Indicators: an application for measuring customer satisfaction [Relazione in Atti di Convegno]
S., Zani; M. A., Milioli; Morlini, Isabella
abstract

This paper deals with the construction of a fuzzy composite indicator ofa latent phenomenon, using a set of manifest variables measured on different scales(quantitative, ordinal and binary). A few criteria for assigning values to the membershipfunction are discussed, as well as for defining the weights of the variables. Forordinal variables we propose a fuzzy quantification method based on the samplingcumulative function and a weight system taking account of the relative frequency ofeach category. An application to obtain a synthetic measure of customer satisfactionfrom the results of a survey is presented.

2010 - La dislessia evolutiva lungo l’arco della scolarità obbligatoria [Capitolo/Saggio]
Stella, Giacomo; E., Savelli; Scorza, Maristella; Morlini, Isabella
abstract

La nostra ricerca vuole aggiornare i risultati ottenuti nel precedente studio (Stella & Cerruti Biondino 2002).Le conclusioni presentate in quel lavoro sulla base dei dati raccolti possono essere così riassunte:• L’evoluzione della lettura nei dislessici italiani nel corso dei 7 anni che intercorrono fra la 2a elementare e la 3a media inferiore risulta migliore per la variabile accuratezza rispetto alla velocità, che evolve molto più lentamente. • La velocità sembra essere la variabile prognostica più attendibile per prevedere la successiva evoluzione delle capacità di lettura.• Sulla base della velocità di decodifica è possibile attribuire un gradiente di severità e identificare due sottogruppi di dislessici che sono stati definiti lievi e severi. • La possibilità di discriminare fra lievi e severi sembra possibile a partire dalla 3a elementare, in quanto precedentemente la bassa variabilità dei valori non consente di discriminare fra lievi e severi.• L’attribuzione del sottotipo lieve-severo in 3a elementare rimane stabile nel corso degli anni fino alla 3a media inferiore, indipendentemente dagli interventi di rieducazione effettuati.• I valori della rapidità rilevati in 3a elementare sembrano costituire un indice di predittività in grado di stabilire con sufficiente affidabilità se un dislessico da adulto sarà un dislessico compensato o persistente.

2009 - Clustering with latent variables [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Besides continuous variables, binary indicators on ICT(Information and Communication Technologies) infrastructures andutilities are usually collected in order to evaluate the qualityof a public company and to define the policy priorities. In thispaper we face the problem of clustering public organizations byassuming that these binary attributes are generated from latentcontinuous variables and by estimating the scores of the latentvariables. In economics, these variables are called utilityfunctions and the assumption is that the binary attributes (whichmay be, for example, the presence or the absence of a publicservice or a public utility) are determined by the crossing of acertain threshold in these functions. To compare the proposedclustering approach with the latent class mixture modelling asimplemented in the Latent Gold package we simulate data from asetting where the true group membership is known. Then, we presenta cluster analysis of the Emilia-Romagna municipalities, based ona set of back office and front office indicators, thatdemonstrates the usefulness of the proposed method as a keysupport for policy makers.

2009 - Computational studies with equivalent degrees of freedoms in neural networks [Articolo su rivista]
S., Ingrassia; Morlini, Isabella
abstract

The notion of equivalent number of degrees of freedom (e.d.f.) has been recently proposed in the context of neural network modeling for small data sets. This quantity is much smaller than the number of the parameters in the network and it does not depend on the number of input variables. In this paper, we present numerical studies on both real and simulated data sets assuring the validity of e.d.f. in a general framework. Results confirm that e.d.f. performs more reliably than the total number W of adaptive parameters - which are usually assumed equal to the degrees of freedom of the model in common statistical softwares - for analyzing and comparing neural models. Numerical studies also point out that e.d.f. works well in estimating the error variance and constructing approximate confidence intervals. We then propose a comparison among some model selection criteria and results show that for neural networks GCV performs slightly better. We finally present a simple forward procedure which can be easily implemented for automatically selecting a neural model with good trade-off between learning error and generalization properties.

2009 - New weighted similarity indexes [Relazione in Atti di Convegno]
Morlini, Isabella; Zani, S.
abstract

Moving from the original work of Zani (1982), in thispaper we propose new weighted similarity indexes for variableswith multiple categories. In contrast to traditional measures,these indexes also consider the frequency of the modalities ofeach attribute in the sample and the possible association betweenpairs of variables. The suitability of the indexes is shownthrough an application in marketing.

2008 - Mixed mode data clustering: an approach based on tetrachoric correlations [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

In this paper we face the problem of clustering mixedmode data by assuming that the observed binary variables aregenerated from latent continuous variables. We perform a principalcomponents analysis on the matrix of tetrachoric correlations andwe estimate the scores of each latent variable and reach a datamatrix with continuous variables to be used in fully Guassianmodels or in the k-means cluster analysis. Results on a simulationstudy and on a real data set are reported

2007 - Comparing approaches for clustering mixed-mode data: an application in marketing research, [Relazione in Atti di Convegno]
Morlini, Isabella; S., Zani
abstract

Practical applications often involve mixtures of categorical and continuousvariables. A variety of algorithms has been proposed for clustering mixed mode data.In this paper we apply different techniques on a marketing research problem. We alsopropose an approach based on the consensus between partitions obtained by consideringseparately each variable or subsets of variables having the same scale. This approach maybe applied to data with many categorical variables and does not impose restrictive assumptionson variable distribution. We then suggest a final fuzzy partition with membershipdegrees obtained as a function of the classes determined by the different methods

2007 - Computational experiences with equivalent degrees of freedoms [Relazione in Atti di Convegno]
S., Ingrassia; Morlini, Isabella
abstract

Neural networks, radial basis functions and projection pursuit regression arenonlinear models which simultaneously project the m-dimensional input data into a p dimensionalspace and model nonlinear functions of the linear combinations of the inputsin this new space. Previous statistical theory for estimating the true error variance andconstructing approximated confidence intervals seems inappropriate, since the degrees offreedom of these models do not equal the number of adaptive parameters. We show inthis article that the problem maybe overcome by using the equivalent degrees of freedom(e.d.f.) based on the dimension of the projection space. We present the results of a MonteCarlo study on simulated data showing that e.d.f. : give numerical stable results and seemto work reasonably well in estimating the error variance and constructing confidence intervals

2007 - Equivalent number of degrees of freedom for neural networks [Relazione in Atti di Convegno]
S., Ingrassia; Morlini, Isabella
abstract

The notion of equivalent number of degrees of freedom (e.d.f.) to be usedin neural network modeling from small datasets has been introduced in Ingrassiaand Morlini (2005). It is much smaller than the total number of parameters andit does not depend on the number of input variables. We generalize our previousresults and discuss the use of the e.d.f. in the general framework of multivariatenonparametric model selection. Through numerical simulations, we also investigatethe behavior of model selection criteria like AIC, GCV and BIC/SBC, when thee.d.f. is used instead of the total number of the adaptive parameters in the model.

2007 - Le reti neurali [Capitolo/Saggio]
Morlini, Isabella
abstract

Il capitolo illustra, a livello didattico, i principali modelli statistici di reti neurali artificiali. Affronta il problema dell'overfitting e mostra le metodologie idonee a stimare un modello con buona capacità di generalizzazione

2007 - Searching for structure in air pollutants concentration measurements [Articolo su rivista]
Morlini, Isabella
abstract

When studying air pollution measurements at different sites in a spatial area, we may search for a typical pattern,common to all curves, describing the underlying air pollution process in a pre-specified period. Another area ofinterest to support local authorities in air quality management may be the classification of the different sites inhomogeneous clusters and the group ranking that follows. Yet, there is variation in both amplitude and dynamicsamong the air pollutant concentrations measured at the different monitoring stations. Analyzing such measurements,where the basic unit of information is the entire observed process rather than a string of numbers, involvesfinding the time shifts or the warping functions among curves. The analysis is much more complicated if weconsider a multivariate process, that is, vector-valued air pollutant measurements. Following our previous workwhere an improved dynamic time-warping algorithm has been developed, especially in the multivariate case, andused both for classifying functional data and estimating the structural mean of a sample of curves, we analyzed themeasurements of some air pollutants in Emilia Romagna (northern Italy). In addition, for the univariate analyses,we applied the self-modeling warping function approach, which is also convenient for these data. Indeed, thismethod was found to be model-free and enough flexible to capture very complex and highly non-linear patterns.

2006 - On multicollinearity and concurvity in some nonlinear multivariate models [Articolo su rivista]
Morlini, Isabella
abstract

Recent developments of multivariate smoothing methods provide a rich collection of feasible models for nonparametric multivariate data analysis. Among the most interpretable are those with smoothed additive terms. Construction of various methods and algorithms for computing the models have been the main concern in literature in this area. Less results are available on the validation of computed fit, instead, and many applications of nonparametric methods end up in computing and comparing the generalized validation error or related indexes. This article reviews the behavior of some of the best known multivariate nonparametric methods, based on subset selection and on projection, when (exact) collinearity or multicollinearity (near collinearity) is present in the input matrix. It shows the possible aliasing effects in computed fits of some selection methods and explores the properties of the projection spaces reached by projection methods in order to help data analysts to select the best model in case of ill conditioned input matrices. Two simulation studies and a real data set application are presented to illustrate further the effects of collinearity or multicollinearity in the fit.

2005 - Estimation of the structural mean of a sample of curves by dynamic time warping [Relazione in Atti di Convegno]
Morlini, Isabella; S., Zani
abstract

Il lavoro estende i risultati sul Dynamic Time Warping (DTW) proposti in Morlini (2004) e Corbellini & Morlini (2004). L’algoritmo DTW viene qui utilizzato per la stima delle funzioni di warping utilizzate nell’allineamento di dati (discreti) campionati da curve e per il successivo calcolo della media strutturale delle serie. La scelta del parametro  nelle splines utilizzate per ottenere una stima smooth dei punti di ogni serie viene effettuata in via automatica, attraverso la cross validation. Si propongono applicazioni sulle quantità medie giornaliere di alcuni inquinanti rilevati dalle stazioni per il monitoraggio della qualità dell’aria in Emilia Romagna.

2005 - L’ICT nella Pubblica Amministrazione: un’ applicazione ai comuni dell’ Emilia Romagna [Relazione in Atti di Convegno]
A., Cerioli; M. A., Milioli; Morlini, Isabella
abstract

In questa nota si affronta il problema della classificazione delle iniziative di e-government intraprese dai comuni dell’Emilia-Romagna. A tale scopo sono analizzate le informazioni rilevate attraverso un questionario ad hoc predisposto dalla Regione Emilia-Romagna.

2005 - Neural network modeling for small datasets [Articolo su rivista]
Ingrassia, S; Morlini, Isabella
abstract

Neural network modeling for small datasets can be justified from a theoretical point of view according to some of Bartlett's results showing that the generalization performance of a multilayer perceptron (MLP) depends more on the L-1 norm parallel to c parallel to(1) of the weights between the hidden layer and the output layer rather than on the total number of weights. In this article we investigate some geometrical properties of MLPs and drawing on linear projection theory, we propose an equivalent number of degrees of freedom to be used in neural model selection criteria like the Akaike information criterion and the Bayes information criterion and in the unbiased estimation of the error variance. This measure proves to be much smaller than the total number of parameters of the network usually adopted, and it does not depend on the number of input variables. Moreover, this concept is compatible with Bartlett's results and with similar ideas long associated with projection-based models and kernel models. Some numerical studies involving both real and simulated datasets are presented and discussed.

2005 - On the dynamic time warping for computing the dissimilarities between curves [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Dynamic time warping (DTW) is a technique for aligning curves that considers two aspects of variations: horizontal and vertical, or domain and range.This alignment is an essential preliminary in many applications before classification or functional data analysis. A problem with DTW is that the algorithm may fail to find the natural alignment of two series since it is mostly influenced by salient features rather than by the overall shape of the sequences. In this paper, we firstdeepen the DTW algorithm, showing relationships and differences with the curve registration technique, and then we propose a modification of the algorithm that considers a smoothed version of the data.

2004 - On the degrees of freedom in richly parameterized models [Relazione in Atti di Convegno]
S., Ingrassia; Morlini, Isabella
abstract

Using richly parameterised models for small datasetscan be justified from a theoretical point of viewaccording to some results due to Bartlett which show that thegeneralization performance of a multi layer perceptron (MLP)depends more on the L1 norm of the weightsbetween the hidden and the output layer rather than on thenumber of parameters in the model.In this paper we investigate the problem of measuring the generalizationperformance and the complexity of richly parameterised procedures and,drawing on linear model theory, we propose a different notion of degrees of freedomto neural networks and other projection tools. This notion is compatible withsimilar ideas long associated with smoothers based models(like projection pursuit regression) and can be interpreted using the projectiontheory of linear models and showing some geometrical properties of neural networks.Results in this study lead to corrections insome goodness-of-fit statistics like AIC, BIC/SBC:the number of degrees of freedom in theseindexes are set equal to the dimension p of the projection spaceintrinsically found by the mapping function.An empirical study is presented in order toillustrate the behavior of the valuesof some selection model criteria.

2004 - Reti neurali e data mining per l’analisi della customer satisfaction: il caso della qualità della didattica nella Facoltà di Economia di Parma [Capitolo/Saggio]
Morlini, Isabella
abstract

The first aim of this paper is to present an original proposal on the measurement of customer satisfaction. The statistical method is based on two different types of neural networks: the self organizing maps and the radial basis function networks. The latter are implemented with orthogonal least squares selection of the basis functions, in order to avoid unstable results and computational problems in the subset selection algorithm due to multicollinearity in the input matrix. Orthogonal least squares are also much faster than forward selection in data mining, when the input matrix is large. Particularly attention is paid on the pre-processing and processing step (choice of the parameters and stability of the results) and on the definition of multivariate outliers. The case study presented throughout the paper is the measurement of the student satisfaction in the Faculty of Economics of the University of Parma and the quantification of the lag between perceived and expected quality.

2004 - Searching for the optimal smoothing before applying Dynamic Time Warping [Relazione in Atti di Convegno]
A., Corbellini; Morlini, Isabella
abstract

Il lavoro estende i risultati del calcolo delle dissimilarità tra curve proposto in Morlini (2003) e considera la validità dell'utilizzo di stime smooth al posto dei dati originari nel calcolo del dynamic time warping cost (DTWC). La scelta del parametro di smoothing nelle splines utilizzate per conseguire una stima smooth dei punti di ogni serie viene effettuata in via automatica attraverso la cross validation. Si propone un’applicazione sulle quantità di biossido di azoto rilevate dalle 85 stazioni per il monitoraggio della qualità dell’aria in Emilia Romagna. Il programma in Matlab per l’adattamento delle splines alle serie e il calcolo del DTWC può essere richiesto via e-mail al primo autore.

2003 - On the criterion of dynamic time warping for computing the dissimilarity between time series [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Nel presente studio si propone un criterio per il calcolo della dissimilarità tra serie storiche - anche di diversa lunghezza - che considera, oltre ai singoli valori, sia la forma della sequenza dei punti, sia la variabilità sull’asse temporale dovuta al possibile diverso allineamento delle serie. Una semplice distanza, ad esempio una metrica della classe di Minkowki, non è in grado di riconoscere come simili due serie con andamento identico ma non allineate sull’asse temporale. D’altra parte, se i valori sono affetti da “rumore”, il semplice time warping tende ad attribuire tutte le differenze fra i punti allo slittamento sull’asse dei tempi. L’algoritmo proposto ottiene in prima analisi una stima smooth dei punti di ogni serie - attraverso splines - e poi effettua il warping su questi valori. Viene mostrata un’applicazione su carte di controllo simulate.

2003 - Orthogonal Radial Basis Function Selection: An Application in Data Mining [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Il presente studio riguarda il problema della selezione delle funzioni di base radiali in un modello additivo. La procedura comunemente usata, la forward selection, non è particolarmente efficiente nel caso in cui il numero di unità statistiche e la dimensione dello spazio di input siano particolarmente elevati, essendo lenta e complessa dal punto di vista computazionale. Un secondo problema riguarda la multicollinearità nella matrice dei dati. Nel caso in cui le funzioni kernel siano gaussiane, a seconda dell’ampiezza e dei vettori scelti come centri, la matrice delle funzioni di base può risultare anch’essa multicollineare, implicando problemi computazionali nel determinare i coefficienti associati ad ogni funzione. Si propongono come alternativa i minimi quadrati ortogonali e si mostra un’applicazione su dati reali

2002 - Analisi della forma di distribuzione [Capitolo/Saggio]
Morlini, Isabella
abstract

Il capitolo è una descrizione, a livello didattico, degli indici statistici e dei grafici, come il boxplot, utilizzati per l'analisi della forma di distribuzione di una variabile quantitativa.

2002 - Apprendimento in due fasi per reti RBF [Capitolo/Saggio]
Morlini, Isabella
abstract

Nel capitolo vengono illustrati i seguenti metodi di stima dei parametri di una rete neurale a funzioni base radiali: la forward selection, i minimi quadrati ortogonali (con e senza parametro di regolarizzazione), l'algoritmo EM.

2002 - Facing multicollinearity in data mining [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Il presente studio riguarda il problema della scelta di un modello di regressione non lineare che si presenta nel data mining quando la funzione che lega una variabile dipendente ad un pluralità di variabili esplicative non è nota ma deve essere desunta dai dati. Viene mostrato come, in presenza di multicollinearità, la scelta del modello non possa essere basata unicamente sull’errore quadratico od indici ad esso collegati (ad esempio, AIC, BIC/SBC), in quanto alcuni modelli che utilizzano l’algoritmo di backfitting sono soggetti a grande instabilità ed arbitrarietà nella scelta delle funzioni di base. Il comportamento dei più noti metodi non lineari basati sia sulla subset selection sia sulla proiezione delle variabili, in presenta di multicollinearità, viene illustrato attraverso un esempio numerico.

2002 - Le reti con funzioni a base radiale [Capitolo/Saggio]
Morlini, Isabella
abstract

Nel capitolo vengono illustrate le caratteristiche di una radial basis function network e sono evidenziati i punti di somiglianza con modelli più noti in statistica come la regressione non parametrica kernel, le spline multidimensionali e le componenti principali non lineari. Vengono infine mostrati i campi di applicazione di questi modelli

2002 - Modelli Neuronali per piccoli insiemi di dati [Capitolo/Saggio]
S., Ingrassia; Morlini, Isabella
abstract

Nel lavoro si affronta il problema della costruzione di modelli statistici di tipo predittivo in situzioni in cui si dispone di un numero esiguo di dati e la relazionen di dipendenza che si vuole stimare è fortemente non lineare. Tipiche applicazioni riguardano le misurazioni indirette o i cosiddetti sensori virtuali. In tali casi il numero di predittori è spesso molto elevato e, a causa della struttura non lineare di dipendenza, non è possibile scegliere opportuni sottoinsiemi di variabili o di loro combinaizoni lineari. Nel presente lavoro tale problema viene affrontato mediante reti neurali a partire dai risultati di Bartlett concernenti le proprietà di generalizzazione di tali modelli. Alcuni criteri operativi vengono infine illustrati mediante due casi di studio concernenti il controllo statistico di qualità

2001 - Multivariate analysis of radar images for environmental monitoring [Articolo su rivista]
Morlini, Isabella; Orlandini, Stefano
abstract

The identification of the relationship between radar reflectivity factor Z, expressed in mm6 m-3, and rainfall intensity R, expressed in mm h-1, is crucial for both the calibration and operational phases. The Marshall and Palmer relationship, which links together Z at the lowest constant altitude plan position indicator (CAPPI) level and R, is commonly used in operational hydrology. This relation is of the form Z = aR^(b). Coefficients a and b reflect the dependence of Z from the number and size distribution of meteors present in the volumes scanned by radar beam. However, as a and b (which depend on the type of precipitation) are affected by great variability and both Z and R are affected by errors, detailed statistical analyses of the Z-R relationship may help improving the operational capabilities of weather radar. In this framework, the aim of the paper is twofold: (1) to develop a non-parametric approach which is more flexible and offers more generalisation capabilities than the MP relationship;(2) to use a vector of all 11 reflectivity factors at different CAPPI levels. For this purposes, three kinds of neural networks are developed: the multi-layer perceptron, radial basis function networks and Bayesian networks. Models are trained and tested using a real data set of reflectivity observed by the Monte Grande weather radar (Teolo, Italy) and rainfall intensity measured at five rain gauges in the Cortina d’Ampezzo area (Northern Italian Dolomites), during the June 12, 1997 storm event (from 11.15am to 12.00pm). Reflectivity data are given at 11 CAPPI levels with 15-minute time resolution. Rainfall intensity data are measured at 5-minute time resolution and are averaged over the 15-minute time intervals of radar data to constitute integrated measurements.

2001 - Prediction interval for long financial series [Capitolo/Saggio]
Morlini, Isabella
abstract

This paper develops a method of modelling that may be applied to build approximated prediction intervals for long series with heteroscedastic conditional variance. The first step of modelling is represented by the estimation of the values of the series, given a vector of exogenous or lagged endogenous variables, by means of a flexible non-linear function in which the parameters are selected with the minimisation of the weight-decay cost function. The second step is represented by the estimation of the conditional variance associated with each prediction, through a second non-linear function in which the target values are given by the squared errors obtained in the first step. An application on 4 financial index series illustrates the method and shows its good forecasting ability.

2001 - Using radial basis function networks for classification problems [Capitolo/Saggio]
Morlini, Isabella
abstract

Multi-layer perceptron is now widely used in classification problems, whereas radial basis function networks (RBFNs) appear to be rather less well known. Purpose of this work is to briefly recall RBFNs and to allow a synthesis of theirs best features. The relationships between these networks and other well-developed methodological tools for classification, both in neural computing and in statistics, are shown. The application of these networks to the forensic glass data set, which is not new in literature, try to lay out what is common and what is distinctive in these networks and other competitive methods and to show, through empirical validation, the networks performance.

2000 - A non linear regression model for time series with heteroskedastic conditional variance [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Il lavoro presenta una metodologia per l’applicazione di modelli non lineari all’analisi delle serie storiche con varianza condizionale eteroschedastica. La metodologia è applicata a serie simulate da processi AR(2) con disturbi ARCH(1) e GARCH(1,1).

2000 - Artificial neural network estimation of rainfall intensity from radar observations [Articolo su rivista]
Orlandini, Stefano; Morlini, Isabella
abstract

Artificial neural networks are used to identify the relationship between weather radar observations of the reflectivity factor Z and rain gauge measurements of rainfall intensity R. These networks are trained and tested using a real data set of reflectivity observed by the Monte Grande weather radar (Teolo, Italy) and rainfall intensity measured by five rain gauges in the Cortina d’Ampezzo area (Italian Dolomites). A principal components analysis is also carried out to investigate the correlation between the Z values at 11 constant altitude plan position indicator levels and to synthesise these values into fewer orthogonal input variables for the networks. Besides the widely used Marshall-Palmer relationship, linear models and flexible discriminants like generalised additive models are used as a benchmark against which the predictive performances of the neural models are measured.Volumetric scans of radar reflectivity Z and gage measurements of rainfall intensity R are used to explore the capabilities of three artificial neural networks to identify and reproduce the functional relationship between Z and R. The three networks are a multilayer perceptron, a Bayesian network, and a radial basis function network. For each of them, numerical experiments are conducted incorporating in the network inputs different descriptions of the space-time variability of Z. Space variability refers to the observations of Z along the vertical atmospheric profile, at 11 constant altitude plan position indicator levels, namely ZT=(Z1,...,Z11). Time variability refers to the observations of Z at the time intervals prior to that for which the estimate of R is provided. Space variability is evaluated by performing a principal component analysis over standardized values of Z, namely Z~, and the first two principal components of Z~ (which describe 91% of the original variance) are used to synthesize the elements of Z into fewer orthogonal inputs for the networks. Network predictions significantly improve when the models are trained with the two principal components of Z~ with respect to the case in which only Z1 is used. Increasing the time horizon further improves the performances of the Bayesian network but is found to worsen the performances of the other two networks.

2000 - Neural network identification of Z-R relationships [Relazione in Atti di Convegno]
Orlandini, Stefano; Morlini, Isabella
abstract

Rainfall is one of the most difficult elements of hydrologic cycle tomeasure and forecast. This is due to the tremendous range of variability it displays over a wide range of scales both in space and time. Weather radar constitutes an attractive possibility for improving the description of rainfall fields. Radar emits electromagnetic energy in narrow bands.From the reflected energy that returns to the transmitter it is possible to obtain measurements of the rainfall field. In practice, the exclusive use of radar is yet to be achieved and rain gauge or other punctual systems are required to calibrate radar. The identification of the relationshipbetween radar reflectivity factor and rainfall intensity is crucial for both thecalibration and operational phases. In this framework, the aim ofthe paper is twofold:(1) to develop a non-parametric approach which is more flexible and offersmore generalisation capabilities than the Marshall and Palmer relationship, and(2) to use a vector of all 11 reflectivity factors at different CAPPI levels.For this purposes, three kinds of neural networks (NNs) are developed:the multi-layer perceptron (MLP), radial basis function networks (RBFNs)and Bayesian networks (BNs) Models are trained and tested using a real data set of reflectivity observedby the Monte Grande weather radar (Teolo, Italy) and rainfall intensitymeasured at five rain gauges in the Cortina d'Ampezzo area (NorthernItalian Dolomites), during the June 12, 1997 storm event(from 11.15am to 12.00pm).

2000 - Stima della relazione tra riflettività radar e precipitazione al suolo mediante reti neurali [Relazione in Atti di Convegno]
Orlandini, Stefano; Morlini, Isabella
abstract

Motivated by a real world problem, this study develops a neural network approach to identify and evaluate the relationship between atmospheric radar reflectivity and ground level rainfall intensity. Rainfall is one of the most difficult elements of hydrologic cycle to measure and forecast. This is due to the tremendous range of variability it displays over a wide range of scales both in space and time. Weather radar constitutes an attractive possibility for improving the description of rainfall fields as it can provide high resolution images in space and time of the atmospheric reflectivity over large areas. Radars emit short pulses of energy in the radio-frequency portion of the electromagnetic spectrum, which are focused by the antenna into a narrow beam. From the backscattering energy of the hydrometeors that returns to the transmitter it is possible to obtain estimates of the rainfall field. Radar data are displayed on constant altitude plan position indicators (CAPPIs) levels. The empirical Marshall–Palmer (MP) relationship is normally used in operational hydrology to link together the reflectivity factor Z at the lowest CAPPI level and rainfall intensity R at the ground level. The coefficients reflect the dependence of Z from the number and size distribution of meteors present in the volumes scanned by radar beam. The MP relationship needs a careful preprocessing phase to remove known anomalies in the data, which are due to several factors such as, for example, the distribution of water particles, low level precipitation and low level evaporation. In addition, much noise may affect radar data, owing to radar calibration, signal attenuation, and electromagnetic signal propagation path. The preprocessing stage may limit the many applications which require a real time estimation of the rainfall field. Furthermore, the MP relationship exploits the radar image at the lowest CAPPI level, while the importance of incorporating the entire vertical profile of Z to improve the estimates of R has been recognised in several works.These considerations raise the following statistical problem: perform a fast multivariate analysis of the Z-R relationship with a view to making real time predictions and filtering the effect of noise and bias and the presence of outliers in the data. The aim of the neural network analysis performed in this work is to stress the differences between the univariate and the multivariate approach to the Z-R relationship and to compare neural networks with the classical MP relationship, in the univariate analysis, and with other flexible non linear statistical methods, in the multivariate analyisis. Performances are evaluated by means of an empirical study.

1999 - Radar images for rainfall measurements: a neural network analysis [Relazione in Atti di Convegno]
Morlini, Isabella; Orlandini, Stefano
abstract

1999 - Radial basis function networks with partially classified data [Articolo su rivista]
Morlini, Isabella
abstract

The problem of estimating a classification rule with partially classified observations, which often occurs in biological and ecological modelling, and which is of major interest in pattern recognition, is discussed. Radial basis function networks for classification problems are presented and compared with the discriminant analysis with partially classified data, in situations where some observations in the training set are unclassified. An application on a set of morphometric data obtained from the skulls of 288 specimens of Microtus subterraneus and Microtus multiplex is performed. This example illustrates how the use of both classified and unclassified observations in the estimate of the hidden layer parameters has the potential to greatly improve the network performances.

1999 - Reti Neural Artificiali [Capitolo/Saggio]
Morlini, Isabella
abstract

Nel capitolo viene fornita una presentazione, a livello didattico, dei principali modelli di reti neurali artificiali. Tali modelli sono utilizzati, in campo statistico, sia per la previsione sia per la classificazione e la discriminazione. Particolare attenzione viene posta, nell'esposizione, ai problemi riguardanti l stima dei parametri e la scelta del modello "ottimale".

1999 - Using neural networks to identify the relationship between radar reflectivity and rainfall intensity [Relazione in Atti di Convegno]
Morlini, Isabella; Orlandini, Stefano
abstract

1999 - Using radial basis function networks for classification problems [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Multi-layer perceptron (MLP) is now widely used in classification problems, whereas radial basis function networks (RBFs) appear to be rather less well known. Purpose of this work is to briefly recall RBF networks, and to allow a synthesis of theirs best features. The application of these networks to the forensic glass data set (Ripley, 1996) tries to lay out what is common and what is distinctive in these networks and other well developed methodological tools for classification, and to compare numerical performances.

1999 - Variabilità del benessere economico nelle province dell’Italia settentrionale [Articolo su rivista]
L., Grossi; Morlini, Isabella
abstract

Nel lavoro si presenta una valutazione dei divari economici esistenti fra le province dell'Italia settentrionale, utilizzando una pluralità di indicatori inerenti alle forze di lavoro ed ai livelli di ricchezza e dei consumi. Attraverso il confronto tra il reddito pro capite e le sintesi multivariate degli indicatori, si conferma l'insufficienza del primo quale indice del benessere economico, ampiamente discussa in letteratura. L'applicazione di metodologie comparse recentemente in letteratura (il boxplot bivariato robusto) e di altre tipiche dell'analisi spaziale dei dati (ricerca dei valori anomali spaziali e Trend Surface Analysis) permette di segnalare alcuni aspetti inerenti ai dati economici provinciali non emersi sino ad oggi ed edvidenzia come possa essere fuorviante considerare l'Italia del Nord un'area economicamente omogenea

1998 - Influenza dell’addestramento sulla rappresentazione dei dati nelle reti di Kohonen: un’applicazione [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

L’utilizzo delle reti neurali per la classificazione e la visualizzazione di dati multidimensionali è progressivamente divenuto più frequente, grazie alla disponibilità di calcolatori sempre più veloci e di potenti algoritmi di apprendimento. Il successo delle reti neurali è dovuto soprattutto alla capacità delle stesse di valutare un gran numero di fattori, tollerare problemi di qualità dei dati (come dati mancanti, incompleti o affetti da errore) ed evidenziare patterns non lineari. Le reti, come le metodologie statistiche tradizionali, non sono tuttavia esenti dalle difficoltà previsive derivanti dalla sovraparametrizzazione, e pongono problemi inerenti alla scelta del modello e delle variabili da utilizzare. Per questo, più che allo sviluppo di nuovi algoritmi di apprendimento, molta attenzione viene ora posta sui criteri di pre-elaborazione dei dati, sulla scelta appropriata dell’architettura di rete da utilizzare e sull’adeguata valutazione dei risultati (Bishop, 1995). Il presente lavoro vuole mostrare come la scelta dei parametri e del programma di addestramento nelle reti di Kohonen (Kohonen, 1989 e 1990) possa influenzare notevolmente i risultati, dando un limite alla capacità di autorganizzazione del modello e mostrando, anche per le reti non supervisionate l’importanza delle scelte del ricercatore nella determinazione dell’output.

1998 - Kohonen networks and the influence of training on data structures [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

In this paper Kohonen feature map is applied to the so-called two-spiral problem. Even if this network is unsupervised, the results indicate that the ability to classify or visualize the data structure depends on the training parameters. The example shows, therefore, that the network self-organization can be limited and the choices of the researcher can strongly affect the network output.

1998 - Multivariate outliers detection with Kohonen networks: an useful tool for routine exploration of large data sets [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

In this article we are considering the exploratory graphical approach to multivariate outliers detection based on Kohonen networks (Kohonen, 1982, 1995). These networks, generally known as self-organising maps (SOM), are able to find interesting low-dimensional projections of high-dimensional data. The utility of the SOM based strategy, especially for Statistical Offices, in controlling the quality of data and finding multidimensional outliers, arises from a number of reasons: it is an easy-to interpret tool for routine exploration of large data set, it can be used in every context, without the specification of an underlying model and it requires very low computational costs.An example on a real data set shows that SOM can be expected to work reasonably well in visualising multivariate outliers. In particularly, outliers identified are in a general agreement with those detected by other well-known statistical procedures such as factor analysis and k-means cluster analysis. The SOM is also shown to be a robust method, since any substantial difference in the qualitative behaviour of the algorithm, due to choice of either alternative neighbourhood functions or differently sized maps, is empirically observed.

1998 - Pre-processing and feature extraction in radial basis functions networks [Relazione in Atti di Convegno]
Morlini, Isabella
abstract

Although neural networks have been born in engineering field, they are actually receiving a lot of attention among the statisticians. As a matter of fact, neural networks can be viewed as computational models very similar to statistical models that can be applied on several types of real data set. In this respect, as it is done in statistics, an important step of feature extraction and data transformation should be added to the learning and prediction phases of a neural network. The aim of this paper is to show the influence of this phase, called pre-processing, in radial basis function networks, on a real data set. The success of these networks, with and without data pre-processing, is measured by the discrimination rule and the generalisation to unobserved pattern. The performances of radial basis functions networks are also compared with the results obtained by the discriminant analysis, on the same data set.

1996 - Caratteristiche strutturali e distribuzione settoriale delle attività produttive nei comuni dell'Emilia Romagna [Capitolo/Saggio]
Morlini, Isabella
abstract

Nel capitolo si considera la struttura delle attività produttive, commerciali e terziarie nei comuni dell'Emilia Romagna.

Università degli studi di Modena e Reggio Emilia

Pubblicazioni