Forecasting guava cultivation: An empirical examination of statistical models for area, production, and productivity in Gujarat

Abstract

The present study was carried out to estimate the trends of area, production and productivity of Gujarat. The time series data on area, production and productivity of Guava for the period 1996-97 to 2015-16 were collected from the Directorate of Horticulture, Gujarat state, Gandhinagar. The data from 1996-97 to 2012-13 have been used for model fitting and remaining for testing the forecast. Different polynomial models (linear, quadratic, and cubic) and time series models (ARIMA) were considered. The statistically most suited polynomial models were selected on the basis of adjusted R2, significant regression coefficients, RMSE values, MAE values and assumptions of residuals (Shapiro-Wilk’s test for normality and Run test for randomness). Appropriate ARIMA models were fitted after judging the time series data for stationarity based on graphically, auto-correlation function and partial auto-correlation function. The statistically model was selected on the basis of various goodness of fit criteria viz., Akaike’s information criteria (AIC), Bayesian information criteria (BIC), RMSE values, MAE values and assumptions of residuals (Shapiro-Wilks test for normality and Box-Ljung test for independence). The result showed that most of the cubic (third degree polynomial model) was found suitable for area, production and productivity of Guava. For Guava crop ARIMA (1, 1, 1), (2, 1, 0) and (1, 1, 0) suitable for area, production and productivity, respectively.