Table 2 Hybrid neural network model for...

Table 2

Hybrid neural network model for crop water estimation algorithm

Algorithm
Data source: https://desagri.gov.in/document-report-category/agriculture-statistics-at-a-glance/ Region of study: India Duration: 1950–2023 Time series length: 74 entries Input Historical crop yield data: X_train historical error data: y_train Test crop yield data: X_test Crop yield data for a specific region: X_new Range of hidden neuron count: neuron_range Range of feedback delay: delay_range Set of training algorithms: alg_set Output Predicted crop yield for the test set: y_pred Estimated CWR for the specific region: cwr_est	Step 3: Error forecasting using GRNN historical error data = actual crop yield – predicted crop yield error_train = y_train - e_nn.predict(X_train) train a GRNN model using historical error data for error forecasting: grnn(error_train) Step 4: Hybrid neural network model prediction For each crop yield dataset in the test set y_pred = e_nn.predict(X_test) use the GRNN model to forecast the expected error in the E-NLAR neural network prediction error_pred = grnn.predict(y_pred) final predicted crop yield y_pred = y_pred + error_pred Step 5: Model evaluation and improvement evaluate the performance of the hybrid ensemble neural network model Analyze the effect of different hyperparameters on the performance of the proposed hybrid ensemble neural network and optimize them to improve the model's accuracy Step 6: Estimation of crop water requirement use the trained model to predict the crop yield for a particular crop in each region based on the historical crop yield data: y_pred = e_nn.predict(X_new) estimate the crop water requirement based on the predicted crop yield. cwr_est = f (y_pred), Where f is a function that calculates CWR based on the predicted crop yield
Step 1: Data Preprocessing Normalize the crop yield data to a common scale if needed Divide the dataset into training, validation, and test sets Step 2: Ensemble of NLAR Neural Network (E-NLARNN) model Initialize an empty list to store the trained TA-NLAR neural networks: nn_list Initialize an empty list to store the RMSE of each TA-NLAR neural network on the validation set: rmse_list for each neuron_count in neuron_range: for each delay in delay_range: for each alg in alg_set Train a TA-NLAR neural network using alg, neuron_count, and delay on the training set: nn Evaluate the performance of the “nn” on the validation set using RMSE: rmse nn_list = Append(nn,nn_list) rmse_list = Append (RMSE (nn_list), rmse_list) Select E-NLARNN model: e_nn = nn_list[argmin(rmse_list)] Train a new E-NLARNN model on the combination of the training and validation sets: e_nn.fit(concat(X_train, X_val))

Algorithm
Data source: https://desagri.gov.in/document-report-category/agriculture-statistics-at-a-glance/Region of study: IndiaDuration: 1950–2023Time series length: 74 entriesInputHistorical crop yield data: X_train historical error data: y_trainTest crop yield data: X_testCrop yield data for a specific region: X_newRange of hidden neuron count: neuron_rangeRange of feedback delay: delay_rangeSet of training algorithms: alg_setOutputPredicted crop yield for the test set: y_predEstimated CWR for the specific region: cwr_est	Step 3: Error forecasting using GRNN historical error data = actual crop yield – predicted crop yield error_train = y_train - e_nn.predict(X_train) train a GRNN model using historical error data for error forecasting: grnn(error_train) Step 4: Hybrid neural network model predictionFor each crop yield dataset in the test set y_pred = e_nn.predict(X_test) use the GRNN model to forecast the expected error in the E-NLAR neural network prediction error_pred = grnn.predict(y_pred) final predicted crop yield y_pred = y_pred + error_pred Step 5: Model evaluation and improvement evaluate the performance of the hybrid ensemble neural network model Analyze the effect of different hyperparameters on the performance of the proposed hybrid ensemble neural network and optimize them to improve the model's accuracy Step 6: Estimation of crop water requirement use the trained model to predict the crop yield for a particular crop in each region based on the historical crop yield data: y_pred = e_nn.predict(X_new) estimate the crop water requirement based on the predicted crop yield. cwr_est = f (y_pred), Where f is a function that calculates CWR based on the predicted crop yield
Step 1: Data Preprocessing Normalize the crop yield data to a common scale if needed Divide the dataset into training, validation, and test sets Step 2: Ensemble of NLAR Neural Network (E-NLARNN) model Initialize an empty list to store the trained TA-NLAR neural networks: nn_list Initialize an empty list to store the RMSE of each TA-NLAR neural network on the validation set: rmse_list for each neuron_count in neuron_range: for each delay in delay_range: for each alg in alg_set Train a TA-NLAR neural network using alg, neuron_count, and delay on the training set: nn Evaluate the performance of the “nn” on the validation set using RMSE: rmse nn_list = Append(nn,nn_list) rmse_list = Append (RMSE (nn_list), rmse_list) Select E-NLARNN model: e_nn = nn_list[argmin(rmse_list)] Train a new E-NLARNN model on the combination of the training and validation sets: e_nn.fit(concat(X_train, X_val))

Algorithm

Data source: https://desagri.gov.in/document-report-category/agriculture-statistics-at-a-glance/Region of study: IndiaDuration: 1950–2023Time series length: 74 entriesInputHistorical crop yield data: X_train historical error data: y_trainTest crop yield data: X_testCrop yield data for a specific region: X_newRange of hidden neuron count: neuron_rangeRange of feedback delay: delay_rangeSet of training algorithms: alg_setOutputPredicted crop yield for the test set: y_predEstimated CWR for the specific region: cwr_est

Step 3: Error forecasting using GRNN

historical error data = actual crop yield – predicted crop yield error_train = y_train - e_nn.predict(X_train)

train a GRNN model using historical error data for error forecasting: grnn(error_train)

Step 4: Hybrid neural network model predictionFor each crop yield dataset in the test set

y_pred = e_nn.predict(X_test)

use the GRNN model to forecast the expected error in the E-NLAR neural network prediction

error_pred = grnn.predict(y_pred)

final predicted crop yield y_pred = y_pred + error_pred

Step 5: Model evaluation and improvement

evaluate the performance of the hybrid ensemble neural network model

Analyze the effect of different hyperparameters on the performance of the proposed hybrid ensemble neural network and optimize them to improve the model's accuracy

Step 6: Estimation of crop water requirement

use the trained model to predict the crop yield for a particular crop in each region based on the historical crop yield data: y_pred = e_nn.predict(X_new)

estimate the crop water requirement based on the predicted crop yield. cwr_est = f (y_pred),

Where f is a function that calculates CWR based on the predicted crop yield

Step 1: Data Preprocessing

Normalize the crop yield data to a common scale if needed

Divide the dataset into training, validation, and test sets

Step 2: Ensemble of NLAR Neural Network (E-NLARNN) model

Initialize an empty list to store the trained TA-NLAR neural networks: nn_list

Initialize an empty list to store the RMSE of each TA-NLAR neural network on the validation set: rmse_list

for each neuron_count in neuron_range: for each delay in delay_range:

for each alg in alg_set

Train a TA-NLAR neural network using alg, neuron_count, and delay on the training set: nn

Evaluate the performance of the “nn” on the validation set using RMSE: rmse

nn_list = Append(nn,nn_list)

rmse_list = Append (RMSE (nn_list), rmse_list)

Select E-NLARNN model: e_nn = nn_list[argmin(rmse_list)]

Train a new E-NLARNN model on the combination of the training and validation sets: e_nn.fit(concat(X_train, X_val))

Source(s): Prepared by the authors

[ViewLarge]

Sharing Unavailable