California housing model

ML model on California Housing Dataset

This project is a machine learning pipeline for predicting California housing prices using the California Housing Dataset. It includes data preprocessing, model training, evaluation using cross-validation, and inference — all in a clean and production-friendly script.

Features

• Stratified train/test split using income categories
• Data preprocessing with 'Pipeline' and 'ColumnTransformer'
• One-hot encoding for categorical data
• SimpleImputer to handle missing data
• 'RandomForestRegressor' model
• Cross-validated RMSE evaluation
• Model and pipeline persistence with 'joblib'
• Inference support on unseen data ('input.csv' → 'output.csv')

Download the dataset

• Get "housing.csv" from the official GitHub repository of the Hands-On ML book Place it in the same directory as main_new.py.

Train or Inference

• The script auto-detects if the model is already trained:

If no model exists, it will:

• Load data
• Preprocess
• Train the model
• Save the model & pipeline
• Export test set to input.csv

If model exists, it will:

• Load input.csv
• Predict housing prices
• Save results to output.csv

Dependencies

• Scikit-Learn
• Pandas
• Numpy
• Joblib