A Comparative Study of Few Classifications Techniques

A comparative study of the performance of three classifiers, Logistic regression, Discriminant analysis, Naïve Bayes’ classifier was conducted using the ‘Credit card defaulter’ data. The relative comparison of the classifiers was done using measure of accuracy and precision obtained from the confusion matrix. Cross validation technique was used while constructing the confusion matrix. Study showed that Logistic regression provided better performance based on accuracy measure from the confusion matrix (77.88% accuracy) compared to the other two and the accuracy level of Bayes’ classifier was the least (36.22%). The results of these study are limited to this particular data set and hence cannot be extended as a general result.