Analytical Performance of Traditional Feature Selection Methods on High Dimensionality Data

Abstract

Dimensionality Reduction is a technique to select features or split contents from a dataset which reduces the dimension. Dimensionality Reduction techniques reduce the computational time to train the Machine Learning Model using the selected features to predict an outcome with higher accuracy. Feature Selection is a part of Dimensionality Reduction which reduces the number of features when developing a model for predictions. Wrapper method is used as Sequential Feature Selection to select the features from the dataset which contributes highly towards the accuracy of the model. Breast Cancer dataset, Vehicle Loan dataset and Loan Defaulter dataset have been used to compare four traditional feature selection algorithms. Once the features are selected from each of the four algorithms, we train the Logistic [15] Regression Model (ML Model) with those features which gives us the computational time and accuracy. Using computational time and accuracy given by the model, of the features selected, of all four algorithms; we put together a comparison.