Browsing by Author "Goyal, Abha"
Now showing 1 - 1 of 1
Results Per Page
Sort Options
ThesisItem Open Access EVALUATION OF ANN (ARTIFICIAL NEURAL NETWORK) AND PENALISED REGRESSION MODELS FOR PREDICTION OF SUNFLOWER (HELIANTHUS ANNUUS L.)YIELD BASED ON WEATHER PARAMETERS IN INDIA(Indira Gandhi Krishi Vishwavidyalaya, Raipur (C.G.), 2020) Goyal, Abha; Shukla, Sindhu; Alivelu, K.; Lakhera, M.L.; Choudhary, V.K.; Awasthi, H.K.India is amongst the largest growers in the oil industry and an importer of edible oil as well. The vegetable oil economy of India is the fourth one in the world after the USA, China, and Brazil. Sunflower is amongst the largest vegetable oil source in the World. Sunflower seeds contain about 48 - 53% of edible oil. Comparatively,sunflower oil is considered to be premium to other vegetable oils. It is used to produce hydrogenated oil and in cosmetic and medical treatments as well. The effect of weather parameters plays an important role in agricultural production such as on plant growth, development, and harvesting. Changes in weather parameters do affect the production of sunflower seeds. Forecasting of sunflower yield is used to analyze past and present behaviors to predict a future oil production that does assist in decision-making and planning for the future ., effectively as well as efficiently. In the present analysis evaluation of artificial neural network (ANN) and penalized regression model for sunflower yield prediction based on the weather parameters in India has been used. The base of our study was the secondary data on the total yield of sunflower crop all over India for 48 years from 1970-2017-18 {collected from www.indiaagristat.com} was undertaken. The meteorological data on five weather parameters (Max Temperature and Min Temperature (degrees in Celsius), rainfall(mm), RH1(%), and RH2(%)) were collected from the India water portal and Indian Meteorological Department (IMD) Pune. In this study, two models have been developed and evaluated for sunflower yield prediction, namely Artificial Neural Network(ANN) and Penalized Regression Model(Ridge regression and LASSO regression). The evaluation criteria for the model are highest R2, lowest RMSE, MAE, and MAPE. Among all the evaluated models, the LASSO regression model has the highest R2 (0.686006), and lowest RMSE(94.07183), MAE(68.37049), and MAPE(11.6727) compared to the remaining models. That implicates, for sunflower yield prediction, the LASSO regression model is the best. The reason behind the best performance of the LASSO regression model for sunflower yield prediction is the quality of variable selection. LASSO model does shrink the less significant regression coefficients equal to zero. LASSO regression gives results that are simpler and relatively more interpretable.