Comparison of Optimized Random Forest Classifier with K-Nearest Neighbors Classifier to Determine the Factors Influencing Teenagers Alcoholic Consumption

This article aims to assess alcohol consumption among teenagers using a unique approach - the application of a Grid Search-based Optimized Random Forest Classifier, with a comparative analysis against the K-Nearest Neighbors Classifier. The dataset employed is sourced from Kaggle and is named “Student Alcohol Consumption”. The sample size for the study was determined using a pretest of 80% g power, a 0.05% threshold, and a 95% Confidence Interval, resulting in the utilization of 20 samples for each of the two groups in the implementation phase. The outcomes of the study indicate that the Optimized Random Forest Classifier outperforms the K-Nearest Neighbors classifier, achieving an accuracy of 90.75%, compared to 83.15%. An independent sample t-test with a significance level of p=0.004 (p<0.05) suggests a meaningful distinction between the two groups. The findings underscore the effectiveness of the intervention, demonstrating that the novel Grid Searchbased Optimized Random Forest Classifier excels in identifying binge drinking among teenagers, boasting a mean accuracy of 90.75%. In contrast, the K-Nearest Neighbors Classifier yielded a mean accuracy of 83.15%.