AI-based payout forecasts

Exciting insights and valuable insights were gained in a very short time. I thank NOVAZOON and Braincourt for this partnership project and the professional work.

Markus Zachai

Responsible Data Science, GEMA

Project and business model content

NOVAZOON & Braincourt provided the joint project team. The following project content was processed:

  • Definition and prioritization of use cases
  • Creation of a benchmark model for the prediction of future distribution amounts of the members
  • Merging different source systems into an interpretable data set
  • Preparation of the data for the optimal form for use in machine learning algorithms
  • Exploratory data analysis and identification of the challenges
  • Template for future machine learning projects, implemented in Jupyter Notebooks with Python

    customer benefits

    On the basis of the project, a distribution forecast was set up and the following content was implemented for GEMA:

    • Feasibility study on predicting payout amounts and recommendations for action to improve results
    • Deeper insights into the data situation and the distribution of distributions to the members, along with identifying the problems that still need to be solved
    • Benchmark model trained with regression algorithms (gradient boosting, random forest, lasso, linear regression and ensemble model)
    • Interpretable results and reusable scripts for training machine learning algorithms