YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
🏡 California Housing Price Prediction
An end-to-end machine learning project that predicts median house values in California using a complete preprocessing and modeling pipeline.
🚀 Project Overview
This project builds a California Housing Price Prediction Model using the classic California Housing Dataset. It includes data preprocessing, feature engineering, model training, evaluation, and automated CSV-based predictions.
The model uses Random Forest Regression and a fully designed Scikit-Learn Pipeline to ensure reproducibility and clean workflow.
📁 Features
✔ End-to-End Machine Learning Pipeline
Handles missing values
Scales numerical features
One-hot encodes categorical features
Uses ColumnTransformer + Pipeline
Performs stratified train-test splitting
✔ Model Training
Trained using RandomForestRegressor
Saves both the preprocessing pipeline and model (pipeline.pkl, model.pkl)
Easily reusable for new predictions
✔ Automated CSV Prediction
Input: input.csv with housing features
Output: output.csv with predicted median_house_value
No manual processing required — the pipeline handles everything
🧰 Tech Stack
Python
Pandas, NumPy
Scikit-Learn
StratifiedShuffleSplit
SimpleImputer, StandardScaler, OneHotEncoder
Pipeline, ColumnTransformer
RandomForestRegressor
Joblib for model persistence
📊 How It Works
1️⃣ Preprocessing Pipeline
The project builds separate pipelines for numerical and categorical attributes:
Numerical → Imputation + Scaling
Categorical → One-Hot Encoding
Combined using ColumnTransformer.
2️⃣ Model Training
Once the data is processed, a Random Forest Regressor is trained to predict the median house values.
3️⃣ Saving the Model
Both the preprocessing pipeline and trained model are saved using Joblib.
4️⃣ Making Predictions
Run the script to process any input.csv and generate output.csv automatically.
🎯 Key Learnings
Building production-style ML pipelines
Handling mixed (numerical + categorical) data
Feature engineering & stratified sampling
Training, tuning, and saving ML models