Application of Logistic Regression on Heart Disease Data and a ]Review of Some Standardization Methods

The purpose of this study is to do a review of logistic regression and its applications. In addition to the review, a comparison of four different methods of standardization of the β - coefficients was done using publicly available Heart Disease Data. The methods were compared using their performance in testing accuracy, training accuracy, and area under the curve (AUC). Based on the comparisons, it was evident that standardizing the coefficient did not affect the overall prediction accuracy of the model regardless of the method used.  Although there was some difference found in the training and testing accuracies, the AUC's were similar to the unstandardized model for all methods. In essence, standardizing facilitates better interpretation and does not affect the predictive accuracy of the model.