Fine-Tuning Prophet model for application in Agro-Industry to forecast Pesticide consumption

Forecasting plays a key role in several industries to predict the probable requirement of a certain product or service over a period of time. This may or may not be dependent on certain external factors depending upon which, a suitable method needs to be chosen to perform the forecasting task. For a given data distribution, the best-fit model needs to be figured out after fine-tuning its hyperparameters to acquire the most optimal forecasts. In this paper, we have used the Prophet model of Facebook, an additive regression model to forecast the requirement of pesticides to be used for agricultural needs, and presented an overview of how fine-tuning the parameters leads to optimal curve fits, hence resulting in better-quality forecasts. We have also studied the impact of the size of the train data on the overall quality of predictions. Hence after combining both the data size and parameter values, the curve fits and predictions have been visually illustrated.