Time Series Forecasting Kaggle Competition
CS639 time series forecasting project and Kaggle competition, ranked first.
Time Series Forecasting Kaggle Competition
This CS639 project centered on time series forecasting through a Kaggle competition. The work covered exploratory data analysis, feature engineering, statistical baselines, and gradient boosted tree methods.
Highlight
Ranked first on the Kaggle competition leaderboard.
What It Does
- Performs EDA on sales, transactions, oil-price, holiday, and store/product-family features.
- Builds forecasting baselines with exponential moving average and ARIMA/SARIMA.
- Trains and compares gradient boosting models including LightGBM, CatBoost, and XGBoost.
- Produces competition-ready submissions evaluated with RMSLE.
Tech Stack
Python, pandas, Plotly, statsmodels, LightGBM, CatBoost, XGBoost, Kaggle Notebook.