FORECASTING WHEAT YIELD IN PAKISTAN (2023-2030), WITH STATISTICAL AND MACHINE LEARNING MODELS

Abstract

Wheat is one of the most important crops in Pakistan, it plays a major role in feeding the population and supporting the economy. This study focusses on predicting wheat production in Pakistan from 2023 to 2030 using four models: Autoregressive Integrated Moving Average (ARIMA), Exponential Smoothing(ES), Trigonometric Box-Cox ARIMA Trend Seasonal(TBATS) and Artificial Neural Network(ANN). We used wheat production data from 1960 to 2023 and analyzed it using the R programming language. The performance of each model was tested using common error measures like RMSE, MAE and MAPE. Among all models TBATS give most accurate results by handling seasonal and trends patterns more effectively. The results show wheat production in Pakistan is likely to increase over next seven years. This forecast can help government and farmers plan better for future food needs and agriculture development.