You’ll receive a folder of Excel and CSV files containing customer-behavior records across several seasons. Your mission is to turn that raw data into a working, well-documented Python pipeline that: • Cleans and unifies the datasets—handle missing values, date formats, duplicate rows, and obvious outliers. • Engineers features that help reveal seasonality (e.g., month, week number, holidays, customer segment). • Trains a machine-learning model that forecasts seasonal sales trends. You’re free to choose an approach—Prophet, XGBoost, LSTM, or another proven method—as long as the code is transparent and reproducible. • Exports predictions to a CSV and generates bar-graph visualizations that compare historical data with forecasted values, including confidence intervals. • Packages everything inside a Jupyter notebook or .py script, backed by a concise README, inline comments, and a requirements.txt so the model can be refreshed with new data in minutes. Acceptance checklist – Code runs end-to-end on a clean Python environment (preferably 3.10+). – Bar graphs render without manual tweaks. – All key steps (prep, training, evaluation, visualization) are clearly explained. If you’re comfortable wrangling data with Pandas, crafting models with scikit-learn or TensorFlow, and presenting results with Matplotlib/Seaborn/Plotly, this project should feel right at home. I’m happy to review milestones as you progress and will provide prompt feedback on the prototype before final delivery.