First of all, thank you for your excellent work make Machine learning job so much easier. I have developed a web application that utilizes the pycaret.
The purpose of this project is to make classification and regression problems as simple as possible.
If you are interested in this project, you can check my repo:
or you can try it out on Streamlit sharing:
The main features are:
- upload CSV or excel as the dataset for training
- simpple EDA
- Automated preprocessing and training
- Multiple result visualization (includes SHAP)
- Online or batch predictions
- Download the whole pipleline for future use.
I have faced some problems, hope you guys can give me some suggestions:
- I also deployed it in Google Cloud Platform. I tried to use Cloud Build to atutomate the CI/CD process.
But it won’t work.
- It turned out, the Streamlit doesn’t support multiple containers yet, if i create deployment and service for Kubernetes separately, the “Sessionstate” will lose.
- And I can’t use xgboost because Streamlit doesn’t know how to cache it. I posted this problem also here:
- I can’t plot some tree-based SHAP Value in GKE, but on my local machine, they worked pretty well.
Thank you in advance, if you guys can help me to improve this app. I am looking forward to hearing from you!