update the readme files

Examples/Multioutput/MO.py

@@ -36,7 +36,7 @@ def df(name, d):
     return X, y
 
 
-X, y = df(name, d)[0], df(name, d)[1]
+X, y = df(name, d)
 
 
 model_gbnn = GNEGNERegressor(total_nn=200,
@@ -75,7 +75,7 @@ def df(name, d):
                random_state=random_state,
                n_cv_general=3,
                n_cv_intrain=3,
-               title=name + '_GBNN_MO_',
+               title=name + '_GBNN_'
                )
 
     gridsearch(X=X,
@@ -87,5 +87,5 @@ def df(name, d):
                random_state=random_state,
                n_cv_general=3,
                n_cv_intrain=3,
-               title=name + '_NN_',
+               title=name + '_NN_'
                )

Examples/README.md

@@ -1,4 +1,17 @@
-# Dataset implementation
+# Example
+
+This package contains the related experiments to the results of the paper. The comparison had done between two proposed GBNNN-Multioutput and GBNN-SingleOutput and neural networks.
+We considered more than 18 datasets for these experiments, and you can access the dataset from the corresponding directory.
+
+# Multioutput
+
+This folder contains the implementation of the Multioutput experiments with the GBNN-MO and NN model.
+It illustrates one dataset as an example. You can expand it for all the used datasets.
+
+# Singleoutput
+This folder contains the implementation of the Multioutput experiments with the GBNN-SO.
+It illustrates one dataset as an example. You can expand it for all the used datasets.
+## Dataset
 
 This directory contains the example of using 17 multioutput regression datasets. I considered MTR datasets for the experiments. The path to clone the dataset and the number of features and missing values are included. The approaches to deal with the missing values are also mentioned in the txt file.
 

Examples/Singleoutput/Gridsearch.py

@@ -14,7 +14,6 @@ def gridsearch(X, y, model, grid,
 
     cv_results_test = np.zeros((n_cv_general, 1))
     cv_results_generalization = np.zeros((n_cv_general, 1))
-    best_index_time = np.zeros((n_cv_general, 2))
     bestparams = []
     cv_results = []
     pred = np.zeros_like(y)
@@ -59,10 +58,6 @@ def gridsearch(X, y, model, grid,
             'mean_test_score'][grid_search.best_index_]
         cv_results_generalization[cv_i, 0] = grid_search.cv_results_[
             'final_test_error']
-        best_index_time[cv_i, 0] = grid_search.cv_results_[
-            'mean_fit_time'][grid_search.best_index_]
-        best_index_time[cv_i, 1] = grid_search.cv_results_[
-            'mean_score_time'][grid_search.best_index_]
 
     results = {}
     results['Metric'] = [
@@ -75,27 +70,16 @@ def gridsearch(X, y, model, grid,
         cv_results_generalization, axis=0)
     results['Std_generalization_score'] = np.std(
         cv_results_generalization, axis=0)
-    results['mean_fit_time'] = np.mean(
-        best_index_time[:, 0], axis=0)
-    results['std_fit_time'] = np.std(
-        best_index_time[:, 0], axis=0)
-    results['mean_score_time'] = np.mean(
-        best_index_time[:, 1], axis=0)
-    results['std_scoret_time'] = np.std(
-        best_index_time[:, 1], axis=0)
 
     pd.DataFrame(results).to_csv(
         title + scoring_functions + '_Summary.csv')
     pd.DataFrame(cv_results).to_csv(
         title + scoring_functions + '_CV_results.csv')
     pd.DataFrame(bestparams).to_csv(
-        title + scoring_functions + '_Best_Parameters.csv')
-    pd.DataFrame(best_index_time, columns=["Fit_time", "Score_time"]).to_csv(
-        title + 'Best_Index_time.csv')
+        title + scoring_functions + '_Parameters.csv')
 
-    np.savetxt(title + 'predicted_values.csv', pred, delimiter=',')
+    np.savetxt(title + '_predicted_values.csv', pred, delimiter=',')
     rm = {}
     rm['mean '] = np.mean(rmse, axis=0)
     rm['std '] = np.std(rmse, axis=0)
-
-    pd.DataFrame(rm, index=['rmse']).to_csv(title + 'RMSE_' + 'score.csv')
+    pd.Series(rm).to_csv(title + '_RMSE_' + 'score.csv')

Examples/Singleoutput/SO.py

@@ -69,5 +69,5 @@ def df(name, d):
                    random_state=random_state,
                    n_cv_general=3,
                    n_cv_intrain=3,
-                   title=name[:-5] + "target_" + str(i) + '_GBNN_SO_',
+                   title=name[:-5] + "_GBNN_SO_target" + str(i),
                    )

README.md

@@ -1,2 +1,39 @@
-# Multioutput_GBNN
-Multioutput GBNN examples
+# Multivariate Gradient Boosted Neural Network 
+
+# About
+The focus of this package is to provide related examples of the paper's experiments and results. You can reproduce our paper's results with the help of this package.
+
+# install
+To run the examples of this package, you have to install the GBNN package from [here](https://github.com/GAA-UAM/GBNN).
+`pip install GBNN`
+# Citation
+If you are using this package, please cite it as below.
+
+```yaml
+References:
+    Type: article
+    Authors:
+      - Seyedsaman Emami
+      - Gonzalo Martínez-Muñoz
+    Arxiv:
+      - https://arxiv.org/abs/1909.12098
+      -
+    Keywords:
+      - Gradient Boosting
+      - "Neural Network"
+```
+
+
+## Key members of Multivariate Gradient Boosted Neural Network 
+
+* [Gonzalo Martínez-Muñoz](https://github.com/gmarmu)
+* [Seyedsaman Emami](https://github.com/samanemami)
+
+# Version 
+0.0.1
+
+## Updated
+2022-04-24
+
+# Date-released
+2022-01-01