Time Series Forcasting on Store Sales

• Tech Stack: Pandas, Plotly, Scikit-learn, Tensorflow, Python
• Kaggle URL: Project Link

Developed LSTM and XGBoost regression models to predict store sales using time series data. Conducted extensive exploratory data analysis (EDA) to extract periodicity-based features and engineered additional variables such as oil prices, holiday events, and promotions to capture external factors influencing sales. Applied interpolation techniques to address gaps in the dataset, ensuring data continuity for model training.

The models achieved high performance, with the XGBoost model attaining an R-squared value of 0.98 on the training data and 0.82 on the testing data. This approach demonstrated the effectiveness of both models in capturing sales patterns and trends, offering valuable predictions for future sales.