FRAUD APP DETECTION OF GOOGLE PLAYSTORE APPS USING MACHINE LEARNING

Abstract

Along the rise in the various mobile applications which are used in daily life, it's more necessary than ever to stay on top of things to decide which are safe and which don't. It is impossibleto pass judgment. Our system is based on four parameters that include ratings, reviews, in app purchases and Contains ad to predict. Our system compares three models Decision Tree classifier, Logistic Regression and Naïve Bayes. These models were further analyzed on four parameters of F1 score, Recall, Precision and Accuracy. A good F1 score should be greater than 0.7 and are call score greater than 0.5 is considered to be good with higher precision and accuracy. On analysis we found Decision tree model as a good model with accuracy of 85%, F1score of 0.815, Recall value of 0.85 and precision of 0.87