if __name__ == "__main__":
algorithms = {
QUESTConstructor().get_name(): QUESTConstructor(),
GUIDEConstructor().get_name(): GUIDEConstructor(),
CARTConstructor().get_name(): CARTConstructor(),
C45Constructor().get_name(): C45Constructor(),
RFClassification().get_name(): RFClassification(),
XGBClassification().get_name(): XGBClassification()
}
genesim = GENESIM()
inTrees_clf = inTreesClassifier()
NR_FOLDS = 5
for dataset in load_all_datasets():
df = dataset['dataframe']
label_col = dataset['label_col']
feature_cols = dataset['feature_cols']
conf_matrices, avg_nodes, times = {}, {}, {}
for algorithm in algorithms:
conf_matrices[algorithm] = []
avg_nodes[algorithm] = []
times[algorithm] = []
conf_matrices['GENESIM'], avg_nodes['GENESIM'], times[
'GENESIM'] = [], [], []
conf_matrices['ISM'], avg_nodes['ISM'], times['ISM'] = [], [], []
conf_matrices['inTrees'], avg_nodes['inTrees'], times[
'inTrees'] = [], [], []
norm_diagonal_sum = sum([
float(conf_matrix[i][i]) / float(sum(conf_matrix[i]))
for i in range(len(conf_matrix))
])
total_count = np.sum(conf_matrix)
accuracies.append(float(diagonal_sum) / float(total_count))
bal_accuracies.append(float(norm_diagonal_sum) / conf_matrix.shape[0])
return {
'acc': (np.around([np.mean(accuracies)],
4)[0], np.around([np.std(accuracies)], 2)[0]),
'balacc': (np.around([np.mean(bal_accuracies)],
4)[0], np.around([np.std(bal_accuracies)], 2)[0])
}
datasets = load_all_datasets()
NR_FOLDS = 5
measurements = {}
figures = {}
for dataset in datasets:
measurements[dataset['name']] = {}
print dataset['name']
conf_matrices = {
'Imbalanced': [],
'RUS': [],
'Tomek': [],
'ENN': [],
'ROS': [],
'SMOTE': [],
'SMOTE SVM': [],
'SMOTE Tomek': [],