Handwritten Digit Classification Using Support Vector Machine Algorithm Comparing with Logistic Regression for Improving Accuracy

The work's objective is to assess how well the classification techniques of logistic regression and support vector machine predict handwritten digits. A Digit dataset with 985 records is used to test the classification system. Random Forest has presented and constructed a framework for Handwritten Digit prediction in the context of comparing Support Vector Machine and LR classifiers. Using the Decision Tree, the sample size for each group was determined to be 32. With an enrollment ratio of 1, alpha and beta values of 0.05 and 0.5, 95% confidence, and 80% pre-test G power, the sample size was calculated using clinical analysis. In terms of accuracy, the SVM classifier predicts the Handwritten Digit classification on the data set with 98.17%, whereas the Logistic Regression yields 88.23%. Between the two groups, there is a statistically significant difference (p=0.003; p¡0.05) with a 95% confidence interval. SVM is therefore superior to classifiers based on logistic regression. The findings demonstrate that, in terms of accuracy, SVM performs better than Logistic Regression.