A Comparative Analysis of Sensitivity and Specificity of LASSO Logistic Regression and Random Forest Using Sonar Data

Abstract

: In this study we carried out a comparison of the sensitivity and specificity of LASSO Logistic Regression and Random Forest in the binary classification of Sonar observations. Here, we utilized the Sonar dataset available in the mlbench R package, comprising 208 observations and 60 predictor variables, with observations classified into Mine and Rock categories. The data were split into training and testing sets using a stratified 80:20 split. LASSO Logistic Regression was fitted using 10-fold cross-validation to determine the optimal regularization parameter, while Random Forest was implemented using 500 decision trees. The predictive performance of the models was evaluated using confusion matrices, with sensitivity and specificity serving as the primary performance measures. McNemar's test was employed to assess whether the two models produced significantly different classification decisions on the same test observations. The results revealed that Random Forest demonstrated higher sensitivity, while both models achieved identical specificity. McNemar's test revealed that the difference in the paired classification decisions was not statistically significant at the 5% level.