~ 1126 ~    Journal of Pharmacognosy and Phytochemistry 2017; 6(6): 1126-1132 E-ISSN: 2278-4136 P-ISSN: 2349-8234 JPP 2017; 6(6): 1126-1132 Received: 25-09-2017 Accepted: 27-10-2017 A Dash College of Agriculture, Chiplima (OUAT), Odisha, India DS Dhakre Institute of Agriculture, PSB, Visva-bharati, Santiniketan, West Bengal, India D Bhattacharya Institute of Agriculture, PSB, Visva-bharati, Santiniketan, West Bengal, India, Correspondence A Dash College of Agriculture, Chiplima (OUAT), Odisha, India Forecasting of food grain production in Odisha by fitting ARIMA model A Dash, DS Dhakre and D Bhattacharya Abstract Agricultural scenario of a state is reflected by the analysis of its food grain production status. In odisha, food grains share a major portion of the total cropped area in kharif season. and rabi. In kharif season, cereals form the major share of food whereas, in rabi the major share is by pulses. A time series modeling approach (Box-Jenkins’ ARIMA model) has been used in this study to forecast food grain production in Odisha. The order of the best ARIMA model was found to be 2,1,0 (without constant) for kharif food grain production and 1,1,0 (without constant) and for rabi food grain production. The selected best fit models are also validated by using the data which were held up and not used for model building. Further, efforts were made to forecast, as accurate as possible, the future food grain production for kharif and rabi season for a period upto three years by using the best fit model. The forecast results have shown that the food grain production will show a positive growth from 2014-15 to 2016-17 for both kharif and rabi season. Keywords: Forecasting, Production, ARIMA, autocorrelation function, partial autocorrelation function. 1. Introduction Agriculture is the backbone of rural economy and livelihood of Odisha. It provides employment both directly and indirectly to about 64 % of the total workforce. It is the largest private enterprise of the State as almost two-thirds of the population of the state is dependent upon agriculture. So, the development of the state is mainly dependent upon the growth in agriculture sector. The agricultural scenario of the state can be best reflected from the analysis of food grains status. This is evident from the fact that, food grain shares almost 86 % and 64 % of the total cropped area in the state, in kharif and rabi season respectively. The forecasting of food grains is of utmost importance for the sake of proper policy framing. The objective of the study is to develop appropriate ARIMA models for the time series of paddy area and production in Odisha and to make three year forecasts with appropriate prediction interval. 2. Methodology ARIMA (Auto Regressive Integrated Moving Average) model describe the present behaviour of the variable under study in terms of linear relationships with its past values. It is an extrapolation method that requires only historical time series data on the variable under study. ARIMA models are developed basically to forecast the corresponding variable. These models have been developed to forecast the cultivable area, production, and productivity of various crops of Tamil Nadu by (Balanagammal, et al. 2000) [1] . These models are also called Box-Jenkins Models on the basis of these authors’ pioneering work regarding time series forecasting techniques (Box et al. 1994) [2] . The main objective in fitting ARIMA model is to identify the stochastic process of the time series and predict the future values accurately. The main steps for setting of Box-Jenkins forecasting model are: (i) Identification of appropriate ARIMA model (ii) Estimation of the parameters (iii) Diagnostic checking (iv) Forecasting. During identification, more than one model may be found suitable for being fitted. In this case, the estimation of parameters is done for each of the selected model. Basing on significance of the estimated parameters, model fit statistics and diagnostic checking of the selected models, the best fit ARIMA model is selected. Then, cross validation of the best fit model is done. After successful cross validation of the best fit model, it is used for forecasting. Usually short term forecasting is done. Description of different steps involved in fitting the appropriate ARIMA model by in the present study is as folows: