Using weather-based machine learning approach to estimate retail sales and interpret weather factors.

We constructed a model that estimates consumer spending (retail sales), a key determinant of underlying economic trends, based on a machine learning model using weather data as an attribute. We also used our estimate model to analyze the impact of temperatures and rainfall on our sales estimates based on the SHAP values we calculated. Our results confirmed the effectiveness of weather data as an attribute in estimating retail industry food & beverage sales. We also identified patterns where weather has a nonlinear impact on sales. For example, lower rainfall results in higher estimated sales, but a moderate increase in rainfall begins to depress sales. Similarly, during extremely cold periods, temperature depresses sales, but a moderate increase in temperature contributes to higher sales.