This project analyzes a supermarket’s historical sales data to forecast future product sales over a 28-day period. The dataset contains both sales prices and quantities for each product, enabling in-depth exploration of sales patterns and trends.
As part of the team, main contributions were made on:
-
Exploratory Data Analysis (EDA): Conducted in
EDA_1.ipynb, uncovering strong seasonality and trend components that informed the modeling strategy. -
Time Series Modeling in R: Developed and compared four forecasting methods for the supermarket’s top-selling product, FOODS_3_586, in
R_notebook.ipynb. Model performance was evaluated using RMSE, with the aim of identifying the best-fit approach. -
Prophet Forecast Review: With limited involvement in the
forecast_prophet.ipynbnotebook, results were reviewed from the advanced Prophet model implementation.
The final analysis in Sales Report.pdf balances predictive accuracy with business objectives, such as prioritizing total sales predictions over individual product forecasts. The report highlights key findings, methodological strengths, and limitations, offering actionable insights to support strategic decision-making in retail sales planning.