Forecasting Key Retail Performance Indicators Using Interpretable Regression.

Area of Science:

• Business Analytics
• Retail Operations Management
• Predictive Modeling

Background:

• Store performance is crucial for business success.
• Accurate forecasting of key performance indicators (KPIs) aids in efficient operational planning.
• Existing forecasting methods often lack confidence intervals or parameter significance insights.

Purpose of the Study:

• To present a novel regression method for predicting store performance indicators.
• To incorporate easily obtainable data like time of day and day of week into predictions.
• To provide confidence intervals and parameter importance directly from the model.

Main Methods:

• A regression-based predictive model was developed.
• The model utilizes historical data of foot traffic, conversion rate, and sales.
• Easily accessible data such as day of the week and hour of the day are included as features.

Main Results:

• The proposed method achieved comparable performance to existing state-of-the-art techniques.
• Root Mean Square Error (RMSE) values were 0.0713 for foot traffic, 0.0795 for conversion rate, and 0.0757 for sales.
• The method successfully provided confidence intervals and parameter importance without additional steps.

Conclusions:

• The developed regression method offers a robust approach to forecasting key store performance indicators.
• The inclusion of confidence intervals and parameter importance enhances the interpretability and utility of forecasts.
• This method provides valuable insights for retail managers to optimize store operations.