Exploratory Data Analysis of Applications of Encryption for Netflix by ML Models

Abstract

This study examines the performance of various machine learning algorithms in analyzing Netflix's dataset, with a focus on encryption methods used for data protection. We evaluated several models including Ada BoostM1, IBk, Random Forest, and Decision Stump using multiple performance metrics. This work analysis revealed that the Ada BoostM1 model achieved the highest accuracy at 97.14%, outperforming other algorithms. It also demonstrated superior performance in precision (0.97), recall (0.97), and F-Score (0.97). The Random Forest algorithm showed the highest ROC value of 0.98. In contrast, the Decision Stump algorithm consistently underperformed, showing the lowest precision (0.71), recall (0.71), F-Score (0.70), and ROC value (0.64). The IBk model also showed relatively low accuracy at 88.09%. Here evaluated the models using the kappa coefficient and Matthews Correlation Coefficient (MCC). Ada BoostM1 achieved the highest scores in both metrics (0.94 for kappa and MCC), while Decision Stump showed the lowest (0.40 for kappa and 0.41 for MCC). Our findings suggest that Ada BoostM1 and Random Forest algorithms are the most effective for analyzing Netflix's dataset, potentially offering insights into the company's competitive strategy and development model. This research contributes to understanding the application of machine learning in analyzing streaming service data and the effectiveness of various algorithms in this context.