Effect Of Choice Of Forecasting Metric On Profit Margin

Abstract

Success in retail industry is highly correlated with the ability to accurately estimate the future demands. However, the concept of accuracy of future demands is itself unclear and identifying a metric that fits best for the problem in hand is not straightforward. In this work, we study the effect of different forecasting metrics on overall business output, in multiple simulated retail settings. We obtain forecasts for every metric by optimising it over the assumed demand distribution and use this forecast in an inventory management system to calculate overall profit over the simulated period.