Our shopper’s finance group manually creates quarterly budgets for 3 enterprise features throughout eight areas utilizing information dumps from a number of sources. The method lacked flexibility and parameter management leading to inaccurate forecasts.

As well as, the forecast improvement was effort intensive and time consuming – took roughly seven hours per area. With a number of areas and consolidation, this effort consumed tons of of hours each month. The tedious handbook course of included collation of knowledge, pre-processing, outlier dealing with, choosing the right mannequin based mostly on error, creation of validation metrics, and visualizing forecasting outcomes.

Scaling the mannequin throughout new divisions and areas consumed a variety of assets; therefore the corporate sought to automate repetitive duties and implement an automatic forecasting answer.