Penguin Classification - MLOps Assignment

Project Overview

This project builds a machine learning pipeline to classify penguin species based on physical measurements. The project also includes an automated system that fetches new penguin data daily from an API, makes predictions, and updates the results.

Technologies Used

• Python 3.11
• Pandas
• Scikit-learn
• Seaborn
• SQLite (via SQLAlchemy)
• Joblib
• GitHub Actions for automation

Workflow Description

• Data Preparation:
  • Load penguin dataset
  • Normalize the database into ISLANDS and PENGUINS tables
• Model Training:
  • Train a Random Forest Classifier to predict species
  • Save the trained model
• Prediction Script:
  • Fetch new penguin data from the API
  • Predict the species
  • Save the result into predictions/latest_prediction.md
• Automation:
  • A GitHub Action runs daily
  • It fetches new data, predicts, and updates the prediction file

Automated Prediction

• The GitHub Action runs daily at 7:30 AM CET
• It fetches new penguin data, predicts, and updates latest_prediction.md

Output

The latest prediction can be found in: "predictions/latest_prediction.md"

It includes:

• Timestamp when the penguin was spotted
• Prediction timestamp
• Predicted species
• Penguin features used for prediction