Forecasting of rice yield of India through non linear growth models

This research paper entitled “Forecasting of rice yield of India through non linear growth models” is based on secondary data. Data was collected for the years 1963 to 2021 from the official sites of indiastat. The data from 1963 to 2016 were analyzed through R- Software and five years of data 2017 and 2021 were kept for model validation of yield forecasting of rice in India. For forecasting rice yield in India, three different nonlinear models namely Monomolecular, Gompertz and Logistic, were used. All three non-linear models were fitted to data by using Statistical software R. For validation of assumptions of residuals i.e., randomness and normality of residuals, Run’s test and Shapiro wilk’s tests were employed respectively while for goodness of fit and validation of models, Chi-square test and eight steps ahead forecasting were done. For getting best best-fitted models for forecasting rice yield, models are compared by seven different statistics R2, RSS, MAPE, MAE, MSE, RMSE,RSE So, after analysing the data, Logistic model is found better for forecasting of rice yield in India with FE% of 6.25% and 5.02 % for the year 2020 and 2021 respectively. Forecasted rice yield for the years 2023 and 2024, calculated by the logistic model and found 2.43 t/h and 2.67 t/h respectively. Forecasting model of rice yield for india is best fitted model (i.e. Logistic) as below.
Y ̂ =4.0048/ (1 +(4.0048/0.8612-1) *exp(-0.0321*t))