Disclosing fast moving consumer goods demand forecasting predictor using multi linear regression

Demand forecasting accuracy undoubtedly influences a company perfomance. With an accurate forecast, the company will be able to utilize its resources efficiently. In practical, most companies only utilize historical selling data as predictor to forecast their product demand either using qualitative forecasting method or time series. However, in this study on a-fast moving consumer goods (FMCG), i.e., insecticide product, these methods do not give good results as expected. The methods produce Mean Absolute Percentage Errors (MAPEs) above 20%. To provide a more accurate forecasting, this study proposes a Multi Linear Regression (MLR) model that uses predictors including climate, promotion, cannibalization, holiday, product prices, number of retail stores, population, and income. The result shows that the MLR gives the best accurate forecast compare to time series methods and simple linear regressions. Using five predictors, i.e., product price, cannibalism, price disparity, fest day and weather, the proposed MLR model gives more accurate forecast with MAPE 8.66%.