The Robustness of Binary Logistic Regression and Linear Discriminant Analysis for the Classification and Differentiation between Dairy Cows and Buffaloes

摘要

This study was planned to evaluate the performance of linear discriminant analysis (LDA) and binary logistic regression (BLR) for differentiation between Friesian cows and buffaloes on the basis of days in milk (DIM), milk yield per year (kg), days open (DO), calving interval (CI), and age at first calving (AFC, month). Considering the assumptions behind each method, LDA and BLR were compared according to sample size impact and lack of multivariate normality of predictors. A random sample of 1070 cases was selected from the animals being represented by all predictors. The comparison between LDA and BLR was based on the significance of coefficients, classification rate, and area under ROC curve (AUC). Results showed that both methods selected DIM, DO and AFC as the significant (P < 0.01) contributors for data classification. The percentages of correct classification were 67.4% and 67.5%, for LDA and BLR, respectively. Besides, The AUCs were 0.660 and 0.664, for LDA and LR, respectively. Overall, sample size has the same impact on both analyses. However, BLR showed slight superiority for animals being correctly classified. In conclusion, LDA and BLR can be used effectively for classification of dairy cattle breeds, even with violation of normality assumption.