Multiple factor analysis and clustering of a mixture of quantitative, categorical and frequency data

Bécue-Bertaut, Mónica; Pagès, Jacques

doi:10.1016/j.csda.2007.09.023

Cited by 168 publications

(117 citation statements)

References 8 publications

Supporting

Mentioning

102

Contrasting

Unclassified

Order By: Relevance

“…We applied different HC methods (23,24) to the dataset, varying distance measures, linkage functions, feature selection (22,25), and then identifying an optimal partition of the inferred trees (26). For control, a set of random trees and clusters was also created.…”

Section: Methodsmentioning

confidence: 99%

“…Variables with a relative frequency below 1% were excluded. We performed EFA both by means of multiple factor analysis and principal component analysis (22). To facilitate visualization of weightings on dimensions, variables were grouped together in terms of (1) lung function (% predicted FEV 1 , % predicted FVC, FEV 1 /FVC ratio, FEF , bronchodilator reversibility, airway hyperresponsiveness), (2) markers of severity or exacerbation (symptoms within the past 4 wk, number of attacks within the last year, hospitalization), (3) family history, (4) comorbidities and atopy (rhinitis, eczema, sensitization, % eosinophils, total IgE); (5) environmental factors (exposure to tobacco smoke, pet ownership), (6) asthma medication (ICS, LABA, montelukast, and any combination), and (7) general characteristics (age, sex, BMI, age of onset of wheeze).…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Challenges in Identifying Asthma Subgroups Using Unsupervised Statistical Learning Techniques

Prosperi

Şahiner

Belgrave

et al. 2013

Am J Respir Crit Care Med

View full text Add to dashboard Cite

Rationale: Unsupervised statistical learning techniques, such as exploratory factor analysis (EFA) and hierarchical clustering (HC), have been used to identify asthma phenotypes, with partly consistent results. Some of the inconsistency is caused by the variable selection and demographic and clinical differences among study populations. Objectives: To investigate the effects of the choice of statistical method and different preparations of data on the clustering results; and to relate these to disease severity. Methods: Several variants of EFA and HC were applied and compared using various sets of variables and different encodings and transformations within a dataset of 383 children with asthma. Variables included lung function, inflammatory and allergy markers, family history, environmental exposures, and medications. Clusters and original variables were related to asthma severity (logistic regression and Bayesian network analysis). Measurements and Main Results: EFA identified five components (eigenvalues > 1) explaining 35% of the overall variance. Variations of the HC (as linkage-distance functions) did not affect the cluster inference; however, using different variable encodings and transformations did. The derived clusters predicted asthma severity less than the original variables. Prognostic factors of severity were medication usage, current symptoms, lung function, paternal asthma, body mass index, and age of asthma onset. Bayesian networks indicated conditional dependence among variables. Conclusions: The use of different unsupervised statistical learning methods and different variable sets and encodings can lead to multiple and inconsistent subgroupings of asthma, not necessarily correlated with severity. The search for asthma phenotypes needs more careful selection of markers, consistent across different study populations, and more cautious interpretation of results from unsupervised learning.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Challenges in Identifying Asthma Subgroups Using Unsupervised Statistical Learning Techniques

Prosperi

Şahiner

Belgrave

et al. 2013

Am J Respir Crit Care Med

View full text Add to dashboard Cite

show abstract

“…Nevertheless, it enables more in-depth analysis than would have been possible by simple observation (López-Íñiguez & Pozo, 2014b). For each question, we apply a two-step analysis, which we shall describe in detail, because lexicometry has not often been used in research into cultural psychology (for an in-depth description, see Bécue-Bertaut, 2008); & Lebart & Salem, 1998).…”

Section: Instrument and Proceduresmentioning

confidence: 99%

Musical Learning and Teaching Conceptions as Sociocultural Productions in Classical, Flamenco, and Jazz Cultures

Casas‐Mas

Pozo

Scheuer

2015

Journal of Cross-Cultural Psychology

View full text Add to dashboard Cite

This study analyzes the discourse of musicians from three different cultures of musical learning, ranging from the more formal classical European culture, through the jazz culture, to the less formal flamenco culture in Roma communities. It is based on cultural studies of learning and education and the implicit conceptions theory. Thirty-one semi-professional guitarists were interviewed about learning and teaching music. We applied the lexicometrical method using correspondence analysis. We found significant lexical differences among the three cultures for all the three educational dimensions analyzed (teaching, learning, and evaluation). We describe literal answers from the most representative participants from each culture (using the automatic selection of modal response procedure according to χ 2 distance) and a qualitative analysis of their full answers. Finally, we project a distribution of the three cultures of learning onto a factorial plane, which summarizes distribution of the three cultures of learning according to two axes that we have interpreted in terms of (a) locus of control (self-others) and (b) phenomenology (analytical-emotional distance-conceptual-explicit knowledge/sensoryinvolvement-embodied-implicit knowledge), respectively. The discourse of classical and flamenco participants expressed other-regulated learning, although classical participants were closer to an explicit, conceptual pole, whereas flamenco participants were closer to an implicit, embodied pole. The discourse of jazz participants lay in between the other two, closer to the explicit pole, but including characteristic language about self-regulation. Keywords learner, teacher, musical learning, cultural psychology, social cognitionResearch into music over the past 15 years has shown growing interest in the non-formal and the informal realms, with the aim of evaluating how they could be applied to teaching within institu-

show abstract

“…As customer characteristics are numerical and categorical variables, we consider the framework of Factor Analysis of Mixed Data (FAMD) introduced by Escofier in 1979 (see [20] and [21] for further details). This framework provides the distance measure for mixed data that we use in CAPRE.…”

Section: B Actionability Of Counter-examplesmentioning

confidence: 99%

CAPRE: A New Methodology for Product Recommendation Based on Customer Actionability and Profitability

Piton

Blanchard

Guillet

2011

2011 IEEE 11th International Conference on Data Mining Workshops

View full text Add to dashboard Cite

International audienceRecommender systems can apply knowledge discovery techniques to the problem of making product recommendations. This aims to establish a customer loyalty strategy and thus to optimize the customer life time value. In this paper we propose CAPRE, a data-mining based methodology for recommender systems based on the analysis of turnover for customers of speciﬁc products. Contrary to classical recommender systems, CAPRE does not aspire to predict a customer's behavior but to inﬂuence that behavior. By measuring the actionability and proﬁtability of customers, we have the ability to focus on customers that can afford to spend larger sums of money in the target business. CAPRE aggregates rules to extract characteristic purchasing behaviors, and then analyzes the counter-examples to detect the most actionable and proﬁtable customers. We measure the effectiveness of CAPRE by performing a cross-validation on the MovieLens benchmark. The methodology is applied to over 10,000 individual customers and 100,000 products for the customer relationship management of VM Matériaux company, thus assisting the salespersons' objective to increase the customer value

show abstract

Multiple factor analysis and clustering of a mixture of quantitative, categorical and frequency data

Cited by 168 publications

References 8 publications

Challenges in Identifying Asthma Subgroups Using Unsupervised Statistical Learning Techniques

Challenges in Identifying Asthma Subgroups Using Unsupervised Statistical Learning Techniques

Musical Learning and Teaching Conceptions as Sociocultural Productions in Classical, Flamenco, and Jazz Cultures

CAPRE: A New Methodology for Product Recommendation Based on Customer Actionability and Profitability

Contact Info

Product

Resources

About