This project aims to build an effective machine learning solution for Walmart, one of the leading retail stores in the US, to accurately predict sales and demand.
I applied 5(five) Regression Models--> Multiple Linear Regression (MLR), Ridge Regression, Lasso Regression, Elastic-Net Regression, and Polynomial Regression for predicting Sales and evaluated the model scores for comparison (R2, RMSE, RSS and MSE).
- Walmart often faces challenges due to unforeseen demand, leading to stockouts.
- Current machine learning algorithms fail to adequately account for external factors, resulting in inaccuracies.
Scope: Historical sales data from 45 Walmart stores across various regions in the US.
Key Factors Influencing Sales:
-
Holidays & Events: Prominent holidays like Super Bowl, Labor Day, Thanksgiving, and Christmas.
-
Economic Indicators: CPI (Consumer Price Index), Unemployment Index.
-
Markdown Events: Seasonal promotional markdowns preceding major holidays.
-
Holiday Weighting: Weeks including major holidays are weighted five times higher than non-holiday weeks.
-
The dataset is taken from Kaggle (https://www.kaggle.com/datasets/yasserh/walmart-dataset).
- Develop a robust ML algorithm to accurately predict demand and sales.
- Account for factors like economic conditions and promotional markdowns.
- Handle incomplete historical data for holiday weeks effectively.
- Model the effects of markdowns on sales during major holiday weeks.
- Improved accuracy in demand forecasting.
- Enhanced inventory planning and stock availability.
- Better alignment of markdown events with predicted holiday demands.