I build a graph between customers and their attributes, where the node labels are customer and others attributes category. However the attributes was shared by so many customer, hence it is becoming dense multipartite graph. I want to compute graph-based feature, but because of it’s being dense, i don’t really get any valuable value. I also don’t find many example on multipartite graph based feature engineering, most of it is bipartite based. Is using graph for feature engineering is better when it is bipartite?
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| K-partite graph for feature engineering | 0 | 19 | June 30, 2025 | |
| Building an ML model using Graph features! | 1 | 1187 | November 22, 2018 | |
| Feature extraction on a peer to peer network with DeepGL embedding algorithm and finding communities of similar hosts | 3 | 1133 | January 8, 2019 | |
| DAG graph repesented as a JSON file to ML feature vectors | 1 | 782 | April 14, 2019 | |
| About the Graph + AI category | 1 | 1466 | August 16, 2018 |