Housing Price Prediction using Ames Dataset

Using the Ames, Iowa housing dataset from Kaggle, I built an end‑to‑end machine‑learning pipeline in Python to predict residential sale prices. After an initial exploratory analysis—visualizing the sale‑price distribution, quantifying missingness, and plotting correlation heatmaps—I dropped features with >40% missing data and imputed the rest (median for numerics, constant “Missing” for categoricals). I then wrapped median imputation, one‑hot encoding, and feature scaling into a single scikit‑learn Pipeline for reproducible preprocessing.

I trained and tuned three models (Linear, Ridge, Random Forest) with 5‑fold cross‑validation, landing on a Random Forest that achieved R² = 0.89 and RMSE ≈ $28,715. All steps, from raw CSV to final model, are documented in annotated Jupyter notebooks; outputs (predictions, charts, pickled model) are version‑controlled on GitHub. This project highlights my skills in data cleaning, feature engineering, model selection, and delivering production‑ready artifacts.