Abstract
Wheat is considered as one of the most important products in Iran.
Concerning high cultivation area of wheat in Khuzestan, an instrument is required to
process stored data in order to give information resulted from such processing to
managers of agricultural sectors. Data mining technique is able to give essential
information and models to producers of wheat for modelling energy consumption. One
of the most practical algorithms is an artificial neural network. The main aim of this
research is to predict output energy of wheat farms using a multilayer perceptron neural
network. This is an analytic research and its database consists of 1240 records. Data
required for the research was obtained from wheat farm during 2014-2018. Data
analysis was done via IBM SPSS modeller 14.2 and standard CRISP. Concerning the
model used in the research, it was found that variables of chemical fertilizers,
machinery & diesel fuel with coefficients of 0.2987, 0.2064 and 0.1527 respectively
had the highest effect on output variable (productive energy). Amount of prediction
precision in neural network algorithm, meaning ratio of correctly predicted records to
total records was 93.08%. Also, linear correlation between actual values and predicted
values were 0.92 and 0.88 respectively, for training data and testing data suggesting
strong correlation. The results obtained can be effective for wheat farmers in direction
of evaluation and optimization of energy consumption in process of wheat production
and reduction of consumption of energy inputs.
Concerning high cultivation area of wheat in Khuzestan, an instrument is required to
process stored data in order to give information resulted from such processing to
managers of agricultural sectors. Data mining technique is able to give essential
information and models to producers of wheat for modelling energy consumption. One
of the most practical algorithms is an artificial neural network. The main aim of this
research is to predict output energy of wheat farms using a multilayer perceptron neural
network. This is an analytic research and its database consists of 1240 records. Data
required for the research was obtained from wheat farm during 2014-2018. Data
analysis was done via IBM SPSS modeller 14.2 and standard CRISP. Concerning the
model used in the research, it was found that variables of chemical fertilizers,
machinery & diesel fuel with coefficients of 0.2987, 0.2064 and 0.1527 respectively
had the highest effect on output variable (productive energy). Amount of prediction
precision in neural network algorithm, meaning ratio of correctly predicted records to
total records was 93.08%. Also, linear correlation between actual values and predicted
values were 0.92 and 0.88 respectively, for training data and testing data suggesting
strong correlation. The results obtained can be effective for wheat farmers in direction
of evaluation and optimization of energy consumption in process of wheat production
and reduction of consumption of energy inputs.
Keywords
ANN
Data mining
Energy
predict
wheat.