Home » Uncategorized

Detecting In-App Purchase Fraud with Machine Learning

This article contains phrases taken from the machine learning and analysis world. Data scientists and algorithm engineers will feel more comfortable with reading it although it’s targeted at anyone who is interested in some deep data science learnings. It was written by Ella Gati. Ella is fascinated by machine learning and data science and is excited to be making big data valuable.

Hacking applications such as FreedomiAP CrackeriAPFree, etc. allow users to make in-app purchases for free. With these kinds of hacks the player receives the coins, gems, levels or lives they purchased without paying any money. If the game developer did not implement any validation process on the in-app purchases, such as SOOMLA’s fraud protection, the purchases are recorded as real purchases in his system. As a result, the reported revenue may differ greatly from the real revenue (especially in popular games with lots of fraud).

We would like to make reports as accurate as possible, and to be able to communicate to the game developers the real state of their game. We use machine learning and statistical modeling techniques for our solution. 

With help from a few big games in the GROW data network we were able to build a model that classifies each purchase as real or fraud, with a very high level of accuracy.


What you will find in this article:

  • In-app purchase model features
  • Decision trees to the rescue
  • Fraud classification performance
  • Per game model
  • Cross-games classification
  • Results
  • Implications for game developers

To view the original post, click here. For other articles about machine learning, click here.

Top DSC Resources

Follow us on Twitter: @DataScienceCtrl | @AnalyticBridge