FCM-type Switching Regression Model for Mixed Measurement Level Data Based on Alternating Least Squares Method

Abstract

Fuzzy c-Regression Models (FCRM) is a Fuzzy c-Means (FCM) -type switching regression technique that simultaneously performs data clustering and local regression model estimation by using regression errors for clustering criteria. The alternating least squares method handles mixed measurement level data by iteratively quantifying nominal variables into numerical scores so that the scores suit the current model. This paper considers two algorithms for handling mixed measurement level data in FCM-type switching regression based on the alternating least squares method. The first one constructs a single numerical data space for revealing geometrical relationships among data samples while the second one quantifys nominal variables in each cluster for revealing mutual dependencies among numerical and nominal variables.