Risk Management in E-Commerce—A Fraud Study Case Using Acoustic Analysis through Its Complexity

Abstract

This work aimed to develop business intelligence towards fraud detection using buyer-placed information combined with the sound analysis from a confirmation purchase call. We used a dataset of 789 orders in 2018, provided by different e-commerce websites and calls fulfilled from every Brazilian state. Nine acoustic index features were used, through entropy in sound and vibration, summarizing the audio plus 6 extra features related, added by 12 customer features to compose two different classifiers (Logistic Regression and Random Forest). The acoustic indexes were, in fact, capable of providing better accuracy of the models, showing a probability associated with the voice characteristics, helping decision-making in credit card fraud.