Selection of Appropriate Growth Model for Prediction of Sugarcane Area, Production and Productivity of India

Abstract

For meeting sugar demand and planning for import, export and distribution, it needs prior estimates of its production. During past years, frequent import and occasional export from India has been witnessed. Sugar production depends more on sugarcane production, our forecasts have not met the actual targets achieved. The prediction model developed for forecasting would help the government in precise planning to meet the sugar demand and control its prices in the market besides voiding import. The there is a strong need to develop a forecast model for sugar/sugarcane production using time series data based on sound statistically and economically for assessing sugar/sugarcane production. Measuring the growth of any entity over a time period is important for understanding the past behavior and for future planning. Among the agricultural crops sugarcane is a multi-product crop, every fraction of which finds economic use either as a food, fodder, fuel or fiber and plays a major role in the rural economy. India produces more than 26 million tonnes of sugar through more than 516 sugar mills and nearly 45 million people are engaged in the production of cane and sugar. The present study has been carried out to apply various non-linear models to forecasts the area, production and productivity of sugarcane crop in India. All India sugarcane data production, productivity and total area cultivated during the period of 1930–2010 has been used for comparison of models. The parameters of the model have been estimated using non-linear estimation procedure. In the present study along with four popular growth models Logistic, Gompertz and Monomolecular, exponential other models viz., MMF, 3/4 degree polynomial and sinusoidal models have been used and compared for estimation of the parameters. The results of the parameter estimates and goodness of fit measures indicated that among the three models third order polynomial for area, MMF model for production and Sinusoidal model for yield were found among all the models used.