Predicting Forecast of Sugarcane Production in Pakistan

• Published in: Sugar tech : an international journal of sugar crops & related industries Vol. 25; no. 3; pp. 681 - 690
• Main Author: Hussain, Nabeel
• Format: Journal Article
• Language: English
• Published: New Delhi Springer India 01.06.2023; Springer; Springer Nature B.V
• Subjects: Agriculture; Analysis; Autoregressive models; Biomedical and Life Sciences; Crop production; Food processing industry; Forecasting; International economic relations; Life Sciences; Model accuracy; Model testing; Planning; Pricing; Rankings; Research Article; Statistical analysis; Sugar; Sugar industry; Sugarcane; Baluchistan Pakistan; Pakistan; Sugarcane production; Pakistan; ARIMA model; Forecasting; Time series
• ISSN: 0972-1525; 0974-0740
• DOI: 10.1007/s12355-022-01221-4

Sugarcane plays a vital role in the overall economy of Pakistan. Pakistan ranks among the top 10 sugarcane producing countries in the world. Further, Pakistan ranks among the top 10 sugar-consuming countries in the world. Due to the excessive quantity of Gur and seed manufacturing, there has been a great loss of production of refined sugar in Pakistan to meet the monthly national consumption of a country. Prediction of crop production level is necessary for policymakers in precise planning to overcome the sugar shortage and control its prices in the market. Hence, there is a dire need to develop an appropriate forecasting model for sugarcane production using time-series data for predicting future sugarcane production. The aim of this study is to evaluate the appropriate models for forecasting sugarcane production in Pakistan along its 4 provinces, namely Punjab, Sindh, Khyber Pakhtunkhwa and Baluchistan. Sugarcane production data from 1982 to 2016 were used for model structure and 2017 to 2019 was used to test the model. Auto-regressive integrated moving average (ARIMA) method was used for the selection of suitable production model. Lag order selection was carried out using AIC and SC. RMSE, MAE, MAPE and Theils’ U statistic were used to test the fitted models accuracy of the data. Results indicate that forecast accuracy of MAPE criteria ranged from 2.83 to 11.02 per cent. Results show that ARIMA (2, 1, 6), ARIMA (1, 1, 2), ARIMA (4, 1, 8), ARIMA (1, 0, 3), and ARIMA (4, 1, 2) models are most suitable models for forecasting sugarcane production in Punjab, Sindh, KP, Baluchistan and Pakistan.