Building forecasting model for multi varied products using R shiny at Hexagon Geo systems India

Abstract

Vishal worked as an intern at Hexagon Geosystems, India wherein he was assigned a role of operations intern and specified a task to improve forecasting capabilities for a wide portfolio of products. The as-is forecasting techniques were concentrated around straight line and moving average models that give promising results when the product lies in a smooth category. However, the basic underlying problem that Vishal had to face was that 80% of the products fell under the intermittent, Erratic and Lumpy categories which are supposed to have unreliable forecasts when forecasted using the above-mentioned models. Products are categorized under 4 major clusters that significantly determine their forecasting capabilities based on Average Demand Interval and Coefficient of Variation. So, Vishal kicked off his project by categorizing these products and with a detailed exploratory analysis over the data set. For all but the smooth demand profile, forecast accuracy is not a reliable performance metric. It lacks contextual information and, in the end, leads you to miss the big picture. This induces overstock situations or, on the contrary, poor service level, both situations you want to avoid. This is why you should take some time to understand your products’ various demand patterns, step back, and adjust your expectations.

It was crucial to understand products’ demand patterns and optimize time and resource allocation for forecasting more than 500 products at a single click of a button. Vishal was given the task of developing a robust forecasting model that would be capable of forecasting massive number of products at minimum time allocation.

The case briefs upon various forecasting models used and deals with the process and techniques of building a forecasting model for massive number of products using interactive capabilities in R Shiny and Excel as a basic data repository.