This package is a complementary assistant to and on top of networkx. In order to apply the graph algorithms to bipartite graph link prediction, you need to implement techniques by yourself. Therefore, I felt it useful to fulfill this gap for you guys and the following items are some of the implemented techniques:
1. Prediction
- Preferential attachment
- Adamic Adar
- Jaccard
- Common neighbors
- Katz
2. Evaluation
- Splitting train and test sets
- K-Fold cross validation
3. Output
- Exports the results of the mentioned algorithms to the predictions directory in csv format