About

This repo contains the code for my minimalist end-to-end machine learning tutorial.

• mini-ml.ipynb: basic full pipeline with EDA, preprocessing, model tuning/evaluation/selection/persistence
• titanic-streamlit-app.py: machine learning web app via Streamlit
• rapidminer: no-code version using Rapidminer
• mini-ml-automl.ipynb: AutoEDA and AutoML (AutoGluon)
• mini-ml-clearml.ipynb and pytorch_mnist_clearml.py: MLOps (ClearML)

Local Setup

Python 3 required, see my tutorial to setup Python 3: https://bit.ly/2uX6wAX

Clone the repo, go to the repo folder in Terminal, setup the virtual environment and install the required packages as follows:

$ python3 -m venv venv
$ source venv/bin/activate
$ pip install -r requirements.txt

Run $ jupyter lab or $ code . (if use VSCode) to go over the notebooks.

Kaggle Notebooks

You can also directly run notebooks using Kaggle:

• mini-ml.ipynb: https://www.kaggle.com/harrywang/minimalist-machine-learning-tutorial
• mini-ml-automl.ipynb: https://www.kaggle.com/harrywang/autoeda-and-automl

Machine Learning Web App

Steps to get the streamlit app running (make sure to use the Terminal and the virtual environment is activated):

1. Get the data and model files ready
2. Create a notebook to do analysis and prediction
3. Create an app python file based on the notebook, such as titanic-streamlit-app.py
4. Run the app locally (Local URL: http://localhost:8501) using terminal: streamlit run titanic-streamlit-app.py
5. Stop the app by using ctrl + C or closing the terminal
6. Deploy the app to the cloud for public access via services such as streamlit share, heroku, aws by following my tutorial at https://github.com/harrywang/streamlit-basics. you can see an example at: https://st-demo-harrywang.herokuapp.com/