How analytics is used in forecasting

Abstract

Over the last decade, the science of forecasting has adopted the tools of the data scientist. Prediction today combines traditional demand planning models with the standard tools of machine learning. The result is much improved accuracy over the short term and an enhanced ability to account for the effects of major changes in the economic environment. On the flipside, researchers must now sort through much greater volumes of data in order to identify what might be useful to produce accurate forecasts. The application of machine learning solves what could be a major stumbling block here. So-called ‘data consolidators’ are now emerging to support forecasters by providing access to previously unknown data as well as the tools for using such data creatively. This paper will demonstrate how data from data consolidators may be used by analytics algorithms to improve the accuracy of forecasts.