Application of Data Transformation Techniques and Train-Test split

Three nonlinear methods are evaluated in this paper to get more accuracy on Wisconsin Breast Cancer Data. The application of two data transformation techniques is explored in this project after removing two highly correlated predictors. Additionally, the different train-test split procedure is also assessed in this project. It is observed that Neural Net achieved more than 99% accuracy while using Train: 80%/Test: 20% on standardized data. At the visual comparison of three models, using the ROC curve, Neural Net performed the best.