The goal of discriminant analysis is to predict group membership from a set of predictors and logistic discrimination allows predicting a discrete outcome such as group membership from a set of variables that may be continuous, discrete and dichotomous. People with diabetes mellitus are discriminated for various factors.

The study ‘Discriminant Analysis and Logistic Discrimination: Application to Diabetes Mellitus Data 2000’ using the cross sectional diabetes data collected from Bangladesh Institute of Research and Rehabilitation for Diabetes, Endocrine and Metabolic Disorders (BIRDEM), aims to identify the various socio-economic, demographic and clinical variables that allows better discrimination between the diabetic and non-diabetic patients and in this way also find the much better classification rule. The application of the discriminant analysis reveals that area, physical exercise, percent body mass index (pbmi), age, heredity, have significant influence for the discrimination between the non-diabetic and diabetic group. In the logistic discrimination, sex, age, area, physical exercise, percent body mass index are found as important risk factors. So, it seems that male patients, who live in urban area, having low weight and height and not doing any physical exercise are more likely to develop the disease than the rural patients involved in hard physical work. The ignorance of genetic endowment of parents to next generation, i.e. heredity does not have effect in the disease development and misleading. The educational level of the patients has no impact in discrimination as well as in the logistic discrimination for developing the disease. Finally, from the two procedures it is found that logistic discrimination classifies the patients in higher percentage in case of binary responses than the discriminant analysis.