We all know that social data are well suited for graph based feature engineering, but there are different types of use cases which can have complex interconnections within the data such as Fraud Detection, Recommendation Systems, Identity Resolution as well as physical systems like molecules, etc.
Graph based features could be an important tool in your feature engineering toolbox to leverage complex interconnections in your data. In this hack session, we will discuss the different types of use-cases where graph features can be used as well as different types of graph-based features that can be created for the different use-cases. We will also do a deep-dive into a few real-world use-cases and demonstrate how to create graph-based features to get a significant boost to the ML models.
Key Takeaways for the Audience
- Understanding of different algorithms to generate graph-based features
- Understanding of different types of use-cases where graph based features can be leveraged