An Improved Method for the Data Cluster Based Feature Selection and Classification

Abstract

The improved Support Vector Machine (SVM) algorithm and a Random Forest (RF) algorithm technique for the cluster-based feature selection and classification existed and were calculated. It has been calculated going on the Statlog Dataset. Missing values in clinical data is also a major issue faced by researchers. A four-stage missing value prediction model has been developed to handle missing values. The complete process includes Data cleaning, Feature selection, Train Validation, Parameter tuning, Model testing, Evaluating, Final classification, clustering and prediction. Support Vector Machine (SVM) algorithms have been put in for the data classification. It has been put in based on the class labels and more connect the correlations based on the heatmap. Logistic Regression (LR) machine learning algorithm is also used to estimate the association in the middle of a depending on variable and one or more independent variables, other than it is used to formulate a prediction about a categorical variable against a continuous one. Random Forest is a Supervised Machine Learning(ML) Algorithm that is used commonly in Classification and Regression problems. The results illustrate that the proposed system provided improved accuracy with random forest on the Statlog Dataset.