Time-series predictive analytics on milk production: Mysore Region, Karnataka, India

Dairy production analytics is becoming popular in recent times because of its immense help in decision making with respect to prefiguring dairy product supply chain. With the availability of huge data from dairy farming sector and advent of data analysis techniques it is possible to identify the driving causes for ectopic variations in dairy production and for effective management of the same. This article presents an elaborative analysis of milk production data. Four time series predictive models namely, linear, polynomial, logarithmic and exponential have been developed using real data on milk production of eight years involving 96 monthly productions. The data for the models were availed from, Mysore Milk Dairy, Mysuru, Karnataka State, India. The models were developed using R and they were evaluated for their prediction accuracy using mean average error (MAE), root mean square error (RMSE) and Theil's u-statistic. Among the models, logarithmic model emerged to be the best with low MAE and RMSE and with high u-statistic value followed by an exponential model.