Working with Andrew Eaves of Andalus Consulting a weather response model was developed for Waitrose, a major UK supermarket chain with 290 branches stocking 20,000+ different product lines.

Waitrose had a sophisticated automated demand forecasting system which accounted for day of week, time of year, bank holidays and circumstances such as promotions, but wanted to improve on this by incorporating weather forecasts.

Historical data on demand and weather patterns obtained from the nearest weather station to each branch was analysed and significant relationships uncovered. Demand forecasts were updated using multipliers when specific weather patterns were forecast. The impact of the new forecasts on ordering dates and quantities was determined, and costs of lost sales and wastage were assessed using simulation. Clustering techniques were developed for combining branches and lines which shared similar demand responses to weather.

Analysis showed that, for those commodities where sales were affected by weather, lost sales could be reduced by 6% and wastage by 1%, leading to a reduction in costs of around 2%. The supermarket business is a low-margin industry, with the average profit margin typically ranging from 1 to 2 percent so this result was considered significant.