A Comparative Study of Fuzzy Logic Regression and ARIMA Models for Prediction of Gram Production

Abstract

This study examines the comparison of fuzzy logic regression methodology and Autoregressive Integrated Moving Average (ARIMA) models to determine appropriate forecasting model for prediction of yield production of food crops. Due to environmental changes based on multifactor, uncertainty has increased in yield production of different crops around the globe. So, in these circumstances a novel forecasting method should be applied to get precise information before time. Now a days many robust methodologies are being used for forecasting purposes. Fuzzy logic and regression model is one of them, which is capable in conditions of uncertainty. In this study fuzzy time series forecasting model is applied along with traditional forecasting tool on gram production data of Pakistan to retrieve significant model for prediction. Initially, 7 fuzzy intervals are constructed using fuzzy logic computations with second and third-degree relationship and then evaluated the fuzzified values with different regression models. ARIMA model with different orders of p, d & q are formulated based on statistical accuracy measures such as autocorrelation function (ACF) and partial autocorrelation function (PACF). Applied the techniques of Akaike Information Criterion, Bayesian Information Criterion and other accuracy measures to select appropriate model for forecasting of gram production. Overall, models evaluation demonstrate that fuzzy logic and regression model perform well than ARIMA model in forecasting of gram production. The precise information about yield production will help the policy makers to take decision about import export, management, planning and other related issues.