How datarishavrathore/mlops-house-price works
This repository implements a house-price prediction workflow: clean raw CSV data, train and log a random-forest model, then serve predictions through a FastAPI application and static frontend. The sampled code supports the data and model flows shown below. The README is empty, and the frontend files and deployment wiring were not sampled, so their behavior and the model-artifact handoff to serving cannot be verified.
Main entry points: serve.py Prediction API · data-clean-versioning.py CSV cleaner · train.py Model trainer · index.html Static frontend
Components 11
Data preparation
- Raw CSV
- CSV cleaner
data-clean-versioning.py - Processed data
Model training
- Model trainer
train.py - MLflow tracking
- Model artifact store
Prediction serving
- Prediction API
serve.py - Model loader
serve.py - Static frontend
static/index.html
Other components
- Data operator
- House seeker
Connections 11
- Data operator → CSV cleaner: runs cleaning
data-clean-versioning.py - CSV cleaner → Raw CSV: reads CSV
data-clean-versioning.py - CSV cleaner → Processed data: uploads cleaned CSV
data-clean-versioning.py - Processed data → Model trainer: provides training data
train.py - Model trainer → MLflow tracking: logs metrics and model
train.py - MLflow tracking → Model artifact store: stores model artifact
- Model artifact store → Model loader: supplies configured model
- House seeker → Prediction API: requests prediction
serve.py - Prediction API → Model loader: uses loaded model
serve.py - Prediction API → House seeker: returns price
serve.py - Prediction API → Static frontend: serves index
serve.py