Fuzzy linear discriminant analysis for chemical data sets

This paper introduces a fuzzy linear discriminant analysis (FLDA) for crisp chemical data sets with a few overlapping data points. It is an improvement over ordinary linear discriminant analysis (OLDA). Three data sets have been analyzed, and the experimental results indicate that the performance of FLDA is superior to that of OLDA. In particular, FLDA can give more information about data structure than OLDA.