Analysis and Comparison for Innovative Prediction Technique of Breast Cancer Tumor by Linear Discriminant Analysis Algorithm over Support Vector Machine Algorithm with Improved Accuracy

Abstract

Aim: The objective of this study is to use machine learning algorithms to detect the presence of breast cancer tumors and compare accuracy, sensitivity, and precision between the Linear Discriminant Analysis (LDA) and Support vector machine algorithm (SVM). Materials and Methods: The research uses two sets of data from the Wisconsin Breast Cancer dataset, which is obtained from the UCI Machine Learning Repository. Linear Discriminant Analysis (N=20) and Support vector machine (N=20) with sample size in accordance to total sample size calculated using clincalc.com by keeping alpha error-threshold at 0.05, confidence interval at 95%, enrollment ratio as 0:1, and power at 80%. The accuracy, sensitivity, and precision are calculated using MATLAB software. Results: Comparison of accuracy (%), sensitivity (%), and precision (%) are done using SPSS software using independent sample t-test. Linear Discriminant Analysis algorithm results in an accuracy of 88.25% (p<0.001), the sensitivity of 94.68% (p<0.001), and precision of 84.35% (p<0.001). Support vector machine algorithm results in an accuracy of 97.50%, sensitivity of 95.83%, and precision of 100%. Conclusion: Support vector machine algorithm performed significantly better than Linear Discriminant Analysis algorithm with improved accuracy of 97.50% for breast cancer prediction.