In this study, the presence of long memory in the volatility process of weekly jute prices in the Samsi and Gajol markets of the Malda district (West Bengal) for the period of January 2009 to December 2022 has been investigated. For this objective, the ARCH-LM test and Hurst rescaled range (R/S) analysis are used to determine the ARCH effect and long memory in the volatility process for the series, and the results indicate the presence of the ARCH effect and long memory in conditional variance. Accordingly, the GARCH and FIGARCH models have been applied for modelling and forecasting the volatility of the jute prices. The wavelet method has been used to estimate the fractional difference parameter in the FIGARCH model. The AR (1)-GARCH (1, 1) and AR (1)-FIGARCH (1,0.270,1) models for the Samsi market and the AR (1)-GARCH (1,1) and AR (1)-FIGARCH (2,0.284,1) models for the Gajol market are found suitable at the training stage based on their minimum AIC and BIC. The forecasting performance of these models was evaluated in the validation period with the help of Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE) criteria, and the residuals were examined to ensure that the fitted models were adequate. Finally, the AR (1)-FIGARCH (1,0.270,1) and AR (1)-FIGARCH (2,0.284,1) are found to be the best optimal models for forecasting the jute prices in the Samsi and Gajol markets, respectively.

Crop yield forecasting under the present climate change scenario needs an effective model and its parameter that how crop respond to the weather variable. A number of weather based models have been developed to estimate the crop yield for the various crops at block, district and state level. Among the different model statistical model is more popular and commonly used. The current study was undertaken to evaluate the performance of statistical model for rice and jute yield forecast of four different district viz. Cooch Behar, Jalpaiguri, Uttar Dinajpurand and Dakhin Dinajpur. Among the four districts Cooch Behar district found superior for kharif rice yield prediction (1.46% error with RMSE 177.68 kg/ha) whereas in case of jute crop its performance was the best in the Jalpaiguri district (-0.44% error with RMSE 217.50 kg/ha).

Rapeseed-mustard crop is an important oilseed crop in India. District-wise yield prediction is essential for various location specific decision making. The performance of two machine learning models namely Support Vector Regression (SVR) and Artificial Neural Network (ANN) were compared with basic linear regression model for district-wise yield prediction of rapeseed-mustard crop. The study area for the present investigation were Cooch Behar, Malda, Jalpaiguri and Uttar Dinajpur districts of West Bengal. Yearly unweighted and weighted weather indices were calculated from weekly weather parameters. The indices that significantly affecting yield were selected using stepwise regression for fitting the models. The ANN model was fitted using backpropagation algorithm. The optimum number of neurons in hidden layer for ANN were ranging between two to four. The Tangent hyperbolic function was found to be suitable hidden layer activation function. The nonlinear Radial Basis Function kernel was the best kernel for Support Vector Regression. While evaluating the performance of fitted models in both calibration and validation stages, the ANN model was the best fitted model for Cooch Behar and Malda and SVR was the best fitted model for Jalpaiguri and Uttar Dinajpur districts. It was concluded that the machine learning models outperformed multiple linear regression model for district-wise yield prediction of rapeseed-mustard crop.