def experiment():
accuracy = [[], []]
for i in range(2):
for j in range(0, 6):
cost_mode = "gini-index" if i == 0 else "cross-entropy"
acc = main(cost_mode=cost_mode, max_depth=j)
accuracy[i].append(acc)
print(accuracy)
return accuracy
df_train = pd.DataFrame(X_train)
df_train.columns = feature_names
df_train['class'] = y_train
df_train.to_csv(data_file_train)
df_valid = pd.DataFrame(X_valid)
df_valid.columns = feature_names
df_valid['class'] = y_valid
df_valid.to_csv(data_file_valid)
df_test = pd.DataFrame(X_test)
df_test.columns = feature_names
df_test['class'] = y_test
df_test.to_csv(data_file_test)
##### learning the tree and testing
# add command line in debug/run arguments as: https://github.com/ryanmadden/decision-tree
import decision_tree
decision_tree.main()
# the result -> results.csv
from sklearn import metrics
df_result = pd.read_csv(open('results.csv', 'r'))
y_pred = df_result['class'].values
accuracy = metrics.accuracy_score(y_test, y_pred)
print('accuracy of C4.5 tree: %.3f' % accuracy)
print(' - PY131 -')
def main():
decision_tree.main()
neural_network.main()
k_nearest_neighbors.main()
boosted_dtree.main()
svm.main()
def train_decision_tree(X, y, tscv, grid_search=False):
import decision_tree as dtree
dtree.main(X, y, tscv.split(X))
if grid_search:
dtree.grid_search(X, y, tscv.split(X))