Cahiers du CEREMADE

Unité Mixte de Recherche du C.N.R.S. N°7534
 
Abstract : n this paper, we present an exploratory analysis methodology adapted to multivariate temporal data, with an application to a randomized controlled clinical trial designed for comparing non-diabetic chronic renal failure treatments. The progression of chronic renal disease is assessed through both numerical descriptors and qualitative information concerning censored data when regular dialysis has occurred; analysis requires the introduction of a coding technique. Using a Multiple Factorial Analysis, we compare different types of coding, focusing on the interest of fuzzy coding with its capacity to preserve a large amount of the distributional information. When Correspondence Analysis is performed on the array produced by fuzzy coding, we obtain mean paths, according to sub-group characteristics, on factorial mappings. These procedures have resulted in the identification of a difference between the treatments with an interpretation of the effects in subgroups of patients, which results were not obtainable with traditional survival methodology; they also provide some additional insights for designing further studies about treating renal impairment.
 
 
FUZZY CODING FOR MULTIVARIATE DATA ANALYSIS IN SHORT-LENGTH MULTIPLE TEMPORAL SERIES
GOLDFARB Bernard, PARDOUX Catherine
2003-32
30-09-2003

 
Université de PARIS - DAUPHINE
Place du Maréchal de Lattre De Tassigny - 75775 PARIS CEDEX 16 - FRANCE
Téléphone : +33 (0)1 44-05-49-23 - fax : +33 (0)1 44-05-45-99