top of page

Click Traffic Fraud Detection

in Mobile Application Advertisements

Project developed with data provided by TalkingData - Chinese company - to detect fraudulent clicks on mobile app advertisements and predict whether or not a user will download the app after clicking on the ad.

Accuracy.png
Dashboard.png

In this Project developed in R you will find: Feature Engineering, statistical analysis, exploratory analysis with PowerBi through an interactive click journey dashboard of each IP listed by date, time and application downloaded, opportunities for businesses for different industries such as digital marketing directing ads to the times with the most download / sales conversion possibilities, predictive modeling through three different machine learning models (Cart, kNN and Random Forest), finally presenting a Confusion-Matrix with the data predicted by the model and the maximum accuracy obtained (92% accuracy).

bottom of page