Introduction

I would like to share a recent experience I had using the py-script framework to deploy a machine learning model. The purpose of this discussion is to showcase how I trained an XGBoost model for the Iris dataset, saved it in JSON format, and used py-script to deploy the model successfully. I believe this could be a valuable addition to the py-script demos section, and I'd like to gather feedback and suggestions from the community.

The Journey

Here's a brief overview of what I accomplished:

1. Model Training: I used the XGBoost package to train a machine learning model on the Iris dataset. The trained model performed well during testing and was ready for deployment.

2. Saving the Model: To make the model accessible to the deployment process, I saved it in JSON format. The JSON file was included in the project repository.

3. Code Deployment: Leveraging the py-script framework, I created a deployment script. When users enter data and click the submit button, the trained XGBoost model is initialized using the xgboost.Booster and load_model methods.

4. Predictions: Users can now obtain predictions from the model, and the results are displayed to them in real-time.

Purpose of this Discussion

I believe this experience demonstrates the versatility and capabilities of the py-script framework for machine learning model deployment. It's a prime example of how py-script can make deploying ML models seamless and user-friendly.

Seeking Community Input

Before considering this for the py-script demos section, I'd like to open the floor for discussion. I welcome any feedback, suggestions, or questions from the community. Is there anything you'd like to know or discuss about the process, the code, or the potential addition to the py-script demos?

Share Your Thoughts

Your insights and ideas are greatly appreciated. Let's have a productive discussion about this use case and explore how it can contribute to the py-script ecosystem.

I have used py-script in these 2 applications

1. Iris Classification: Github and live demo
2. Loan Prediction: Github and live demo

Looking forward to your feedback and suggestions!