AnacletoLAB / grape

🍇 GRAPE is a Rust/Python Graph Representation Learning library for Predictions and Evaluations
MIT License
502 stars 38 forks source link

Use case regarding Customer Analytics or Community detection? #29

Open stkarlos opened 1 year ago

stkarlos commented 1 year ago

Thanks for that repo. It seems that you have integrated several tools / libraries / approaches under Grape's hood. Do you intend to create a tutorial for a customer analytics recommendation?

Thanks in advance.

LucaCappelletti94 commented 1 year ago

Hi @stkarlos! Thank you for your suggestion!

I appreciate your interest in using GRAPE for customer analytics recommendations. I have incorporated several libraries into GRAPE, including PyKEEN and wrapped sklearn models. However, I have also implemented several features and models from scratch. Nevertheless, GRAPE's model interfaces allow easy integration of any tool or library that conforms to a sklearn-like interface.

I am working on a tutorial using the MovieLens datasets, but I have encountered some challenges. The MovieLens datasets have a custom format, which means they are not readily loadable in GRAPE. I am working on a pipeline to convert them to simpler TSVs, but this process is still ongoing. Additionally, once the datasets are converted, I would like to integrate them into GRAPE's graph retrieval functionality. However, doing so would require hosting and redistributing the datasets, which requires explicit permission from MovieLens based on their license. I have requested permission and am waiting for their response.

Do you have any specific suggestions for other datasets or topics you would like to see covered in a tutorial? I have already shared link prediction tutorials such as this one on GCNs, and I'd like to add more specific to what you would be interested in.

Thank you again for your suggestion! I look forward to continuing this conversation and exploring ways to use GRAPE for customer analytics recommendations.

stkarlos commented 1 year ago

Thanks for your answer @LucaCappelletti94.

I attach here some popular Kaggle datasets: