README.md

Using examples

These examples are should to be used with Python 3.10. If you encounter any errors, please check your Python version first.

---

MLflow Diabetes - tracking example

Simple example to test tracking metrics to Tracking server

This notebook utilizes machine learning project packaged as MLflow Projects format, and stored in Github.

To run this example you need to have jupyter notebook / lab instance on your own computer, or in cloud (like cPouta)

Install mlflow with command:

pip install mlflow

Set up your credentials to environment variables like:

export MLFLOW_TRACKING_URI=https://<YOUR_APP_NAME>.rahtiapp.fi
export MLFLOW_TRACKING_USERNAME=your_username
export MLFLOW_TRACKING_PASSWORD=your_password

export MLFLOW_S3_ENDPOINT_URL=https://<YOUR_APP_NAME>-minio.rahtiapp.fi
export AWS_ACCESS_KEY_ID=your_generated_access_key
export AWS_SECRET_ACCESS_KEY=your_generated_secret_key 

or with notebook if you have Conda environment.

After that, you should be able to run 01_MLflow-Diabetes-Tracking.ipynb to test MLflow tracking. New experiment and run should appear in Tracking server containing metrics and artifacts of the run. You can run notebook multiple time with different parameters and compare results in Tracking server UI.

---

MLflow Diabetes - model inference example

Simple example to utilize machine learning model from MLflow Tracking server

This example assumes you have run MLflow Diabetes - tracking example at least once, so you have model tracked.

To run 02_MLflow-Diabetes-Inference.ipynb you need to have jupyter notebook / lab instance on your own computer, or in cloud (like cPouta).