Weather-Based Neural Network, Stepwise Linear and Sparse Regression Approach for Rabi Sorghum Yield Forecasting of Karnataka, India

Abstract

Sorghum is an important dual-purpose crop of India grown for food and fodder. Prevailing weather conditions during the crop growth period determine the yield of sorghum. Hence, the crop yield forecasting models based on weather parameters will be an appropriate option for policymakers and researchers to develop sustainable cropping strategies. In the present study, six multivariate weather-based models viz., least absolute shrinkage and selection operator (LASSO), elastic net (ENET), principal component analysis (PCA) in combination with stepwise multiple linear regression (SMLR), artificial neural network (ANN) alone and in combination with PCA and ridge regression model are examined by fixing 90% of the data for calibration and remaining dataset for validation to forecast rabi sorghum yield for different districts of Karnataka. The R2 and root mean square error (RMSE) during calibration ranged between 0.42 to 0.98 and 30.48 to 304.17 kg ha−1, respectively, without actual evapotranspiration (AET) whereas, these evaluation parameters varied from 0.38 to 0.99 and 19.84 to 308.79 kg ha−1, respectively with AET inclusion. During validation, the RMSE and nRMSE (normalized root mean square error) varied between 88.99 to 1265.03 kg ha−1 and 4.49 to 96.84%, respectively without AET and including AET as one of the weather variable RMSE and nRMSE were 63.48 to 1172.01 kg ha−1 and 4.16 to 92.56%, respectively. The performance of six multivariate models revealed that LASSO was the best model followed by ENET compared to PCA_SMLR, ANN, PCA_ANN and ridge regression models because of reduced overfitting through penalisation of regression coefficient. Thus, it can be concluded that LASSO and ENET weather-based models can be effectively utilized for the district level forecast of sorghum yield.

Modeling the Relationship between Rice Yield and Climate Variables Using Statistical and Machine Learning Techniques

“…e models resulted in MSE < 0.1, which is comparable to previous results but with higher correlation coefficients. e yield functions obtained by applying statistical methods and derived in terms of the climatic factors are given in equations ( 9)- (11). Some climatic factors did not appear in the yield function as their impact was minimal compared to rainfall, temperature, and average wind speed.…”

“…As rice is a primary source of food for more than half the world's population, numerous research approaches were proposed for predicting the rice yield [4]. Similar research studies were conducted to model the relationship between the climatic factors and the yield of some other crops such as barley [5], corn [6], sugar cane [7], citrus [8], tea [9], coconut [10], sorghum [11], maize, and soybean [12]. A multiple number of climatic factors were considered in such research studies for the application of statistical methods and machine learning techniques.…”

Comparison Statistical Rice Yield Prediction with Multiple Weather Parameters

“…Where, m-Week of forecast , Xiw/Xii'-Value of i th /i th weather variable understudy in w th week , r j iw /r j ii'w -Correlation coefficient of de-trended yield with i th weather variable/product of i th and i'th weather variables in w th week Detrending of crop yield: Detrending of yield was done to reduce the nonlinear and nonstationary trend that would cause fluctuation in yield prediction. This trend has to be removed before the computation of basic correlation function in order to improve the performance of the model [5][6]. The simple linear regression model used for the detrending of crop yield was Yt = βo+β1t, where Yt -crop yield at given time, βο & β1-Coefficients.…”

“…Absolute Shrinkage Regression Operator (LASSO): It overcome the drawbacks of ordinary least square (OLS) and ridge regression, through various penalties and retains all predictors. The LASSO model is a regression analysis that does both variable selection and regularisation to improve the statistical model's prediction accuracy and interpretability [6,8,9]. LASSO eliminators are utilised for a consistent regression coefficient and automatic variable selection.…”

“…ENET reduces the impact of different features while not eliminating all of the features. [8,6,4,10].…”