some refactor and prior effect script

BayesianKDEy/commons.py

@@ -32,10 +32,13 @@ multiclass = {
     'fetch_fn': fetch_UCI_multiclass,
     'sample_size': 1000
 }
-multiclass['datasets'].remove('poker_hand')  # random performance
+try:
-multiclass['datasets'].remove('hcv')  # random performance
+    multiclass['datasets'].remove('poker_hand')  # random performance
-multiclass['datasets'].remove('letter')  # many classes
+    multiclass['datasets'].remove('hcv')  # random performance
-multiclass['datasets'].remove('isolet')  # many classes
+    multiclass['datasets'].remove('letter')  # many classes
+    multiclass['datasets'].remove('isolet')  # many classes
+except ValueError:
+    pass
 
 
 # utils

BayesianKDEy/plot_simplex.py

@@ -405,7 +405,7 @@ if __name__ == '__main__':
 
     K = 3
     # alpha = [p] + [(1. - p) / (K - 1)] * (K - 1)
-    alpha = [0.095, 0.246, 0.658]
+    alpha = [0.095, 0.246, 0.658]  # connect-4
     alpha = np.array(alpha)
 
 

BayesianKDEy/prior_effect.py

@@ -12,6 +12,7 @@ from quapy.method.aggregative import ACC, AggregativeQuantifier
 from sklearn.linear_model import LogisticRegression as LR
 from copy import deepcopy as cp
 from tqdm import tqdm
+from full_experiments import model_selection
 
 
 def select_imbalanced_datasets(top_m=5):
@@ -22,7 +23,6 @@ def select_imbalanced_datasets(top_m=5):
         balance = normalized_entropy(data_prev)
         datasets_prevs.append((data_name, balance))
     datasets_prevs.sort(key=lambda x: x[1])
-    print(datasets_prevs)
     data_selected = [data_name for data_name, balance in datasets_prevs[:top_m]]
     return data_selected
 
@@ -110,6 +110,7 @@ def experiment(dataset: Dataset, point_quantifier: AggregativeQuantifier, grid:
 if __name__ == '__main__':
     result_dir = Path('./results/prior_effect')
     selected = select_imbalanced_datasets()
+    print(f'selected datasets={selected}')
     qp.environ['SAMPLE_SIZE'] = multiclass['sample_size']
     for data_name in selected:
         data = multiclass['fetch_fn'](data_name)