Open AccessMustard Yield Prediction using State Space Models
Author(s) -
Ekta Hooda,
B. K. Hooda
Publication year - 2020
Publication title -
current journal of applied science and technology
Language(s) - English
Resource type - Journals
ISSN - 2457-1024
DOI - 10.9734/cjast/2020/v39i4831268
Subject(s) - yield (engineering) , autoregressive integrated moving average , mathematics , regression analysis , econometrics , state space representation , agriculture , regression , statistics , time series , geography , materials science , archaeology , algorithm , metallurgy
Forecasting of agricultural outputs well in advance has always been the focus of numerous researchers due to its direct implications on various areas of the society. This study aims to develop State Space Models (SSMs) with weather as an exogenous input over the commonly used ARIMA and regression analysis for yield prediction for mustard crop in eight districts of Haryana state in India. These models are time-varying parameter models and take into account for changes that are known over time in structure of the framework. SSMs with various kinds of growth trends were tried and model performances were compared using AIC and BIC criteria but the growth trend represented by polynomial splines of order-2 with the weather as an exogenous input was chosen as the most appropriate model for mustard yield prediction in all the eight districts under study. Based on the developed models, post-sample yield predictions for the next three years, i.e. 2016-17 to 2018-19 have been obtained and the deviations from actual values are also calculated which came out to be acceptable in an agricultural setup.