• Printed Journal
• Indexed Journal
• Refereed Journal
• Peer Reviewed Journal

International Journal of Statistics and Applied Mathematics

2019, Vol. 4, Issue 6, Part B

Forecasting monthly sugar cane yields using box-Jenkin’s predictive models in Kenya

Author(s): Lydiah Kwamboka, George Otieno Orwa and Zablon Maua Muga

Abstract: Sugarcane is the main raw material in the production of sugar in Kenya. The supply of sugarcane affects directly the quantity of sugar supplied in the markets. Low supply of sugarcane leads to a decline in the amount of sugar supplied to the markets and vice versa. This creates the need of determining the quantity of sugarcane supplied by the farmers to the industries to facilitate planning. This study employed Box-Jenkins predictive models in forecasting the monthly quantity of sugarcane supplied by farmers to the industries. This study will be useful to the government and sugar industries in planning by forecasting the quantity of sugarcane expected to be supplied by farmers. Secondary data on sugarcane yields was analyzed for trend and seasonal components. Kendall’s Tau test was also conducted and it yielded a significant p-value (0.001) compared to the test level (α) = 0.05. This study detrended the data and seasonal ARIMA model was fitted to the monthly sugarcane data. SARIMA (0,1,1)(0,0,0)12 was identified from a list of SARIMA models because it had the lowest Bayesian Information Criterion (BIC). The parameter was identified and a hypothesis test, based on Ljung-Box test, was conducted to determine if the model fitted the cane data. Ljung-Box statistics = 16.577 < tabulated chi-squared value = 27.59 suggesting that SARIMA (0,1,1)(0,0,0)12 fitted the monthly sugarcane data. The R2 = 0.574 indicating that the Box Jenkins model fitted the data. SARIMA (0,1,1)(0,0,0)12 was used to conduct the monthly forecasts. It was noted that the sugarcane yields increased with time.