GROWTH TREND AND YIELD FORECASTING OF MAIZE IN KARNATAKA AND BIHAR THROUGH AUTO-REGRESSIVE INTEGRATED MOVING AVERAGE (ARIMA) MODEL
Loading...
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
RPCAU, Pusa
Abstract
The present study entitled examine “Growth Trend and Yield Forecasting of Maize in Karnataka and Bihar through Auto Regressive Integrated Moving average (ARIMA) Model” is based on the growth trends and ARIMA models for forecasting Maize yield in Bihar and Karnataka. The data spanning from 1990 to 2021 was gathered from reputable online sources such as the Department of Economics and Statistics and India Agri Stat. To create the maize yield prediction model, data up until 2020 were employed, reserving the subsequent two years' data for validating the forecast model. Additionally, trend analysis and tests for model validity were conducted.
Based on the aforementioned information, it was determined that among a range of models including ARIMA (0,1,1), ARIMA (0,1,2), ARIMA (0,0,1), ARIMA (1,0,0), ARIMA (1,1,1), ARIMA (2,0,0), and ARIMA (2,0,1), the ARIMA (2,0,1)
model offers the best fit for forecasting maize yield in Karnataka. Similarly, for Bihar, the ARIMA (2,0,0) model emerged as the most suitable choice from the selection of models, which also included ARIMA (0,0,1), ARIMA (1,0,1), ARIMA (0,1,1), ARIMA (1,1,1), ARIMA (0,1,2), ARIMA (2,0,0), and ARIMA (2,0,1). The
parameters of these models were calculated and subjected to significance tests.
Various statistical measures were also computed to identify the appropriate and efficient model, involving t-tests and chi-square tests. This is reinforced by the presence of low values for MAPE, MAE, RMSE, and BIC in the prediction of maize yield for both Karnataka and Bihar. Forecasts for the next three years' maize yield were generated using ARIMA models. The outcomes revealed a consistent decline in maize yield, both in Karnataka and Bihar. Selected ARIMA model for forecasting of yield of maize in Karnataka and Bihar are as below:
Zt – Zt-1 = 15.841+0.249(zt-1-zt-2)+0.196(zt-2 - zt-3)-0.19(at-1-at-2)+at (for Karnataka)
Zt – Zt-1 = 22.810+0.29 (zt-1 - zt-2) + 0.20 (zt-2 - zt-3) + at (for Bihar)
In this study, lower and upper limits of the forecasted yield were also calculated with 95% of confidence interval. The forecasts done three years period ahead for the time series data of yield of maize by using the best fitted ARIMA (2,0,1) and ARIMA (2,0,0) models, respectively for Karnataka and Bihar.
Further study was done for the trend analysis and it is found that the trend of area, production and yield of maize in Karnataka is in increasing order whereas in Bihar area is in decreasing order and production and yield shows increasing in Bihar. For accuracy coefficient of determination is calculated. Compound Annual Growth Rates were also calculated and it was found that all are highly significant.
Annual income of the majority of the farmers of the study area are in between
1 lakh to 5 lakh and Age, Caste, Occupation, Education, Family size, Size of operational land holding, Farming experience shows a positive correlation with the farmer’s income. The size of operational land shows highly significant with the dependent variable which is farmers income.