Python produce_model_performance_summaryの例

コード例 #1

ファイル: census_knn.py プロジェクト: eazymac25/cs7641-Project-1

best_model = grid_search.best_estimator_
# Time fitting best model
start = timeit.default_timer()
best_model.fit(x_train, y_train)
end = timeit.default_timer()
print('Time to fit:', end - start)
helpers.log_fit_time('CENSUS_KNN', end - start)

# Plot the learning curve vs train size after finding the best model
helpers.plot_learning_curve_vs_train_size(
    best_model,
    df,
    feature_cols,
    'income_num',
    output_location=
    'census_output/knn_%s_best_model_num_samples_learning_curve.png' %
    weighting,
)

# Predict income with the trained best model
y_pred = best_model.predict(x_test)

helpers.produce_model_performance_summary(
    best_model,
    x_test,
    y_test,
    y_pred,
    output_location='census_output/KNN_%s_summary.txt' % weighting,
    cv=kfold,
    scoring='accuracy')

コード例 #2

ファイル: census_svm.py プロジェクト: eazymac25/cs7641-Project-1

    cv=5
)

grid_search.fit(x_train, y_train)

print(grid_search.best_score_)
print(grid_search.best_params_)

# train the best model
best_model = grid_search.best_estimator_
# Time fitting best model
start = timeit.default_timer()
best_model.fit(x_train, y_train)
end = timeit.default_timer()
print('Time to fit:', end-start)
helpers.log_fit_time('CENSUS_SVM', end-start)

# Predict income with the trained best model
y_pred = best_model.predict(x_test)

helpers.produce_model_performance_summary(
    best_model,
    x_test,
    y_test,
    y_pred,
    output_location='census_output/svm_summary.txt',
    cv=kfold,
    scoring='accuracy',
    grid_search=grid_search
)

コード例 #3

ファイル: wine_neural_network.py プロジェクト: eazymac25/cs7641-Project-1

                                                      (20, 5)]
                           },
                           cv=3)

grid_search.fit(x_train, y_train)

print(grid_search.best_score_)
print(grid_search.best_params_)

# train the best model
best_model = grid_search.best_estimator_
# Time fitting best model
start = timeit.default_timer()
best_model.fit(x_train, y_train)
end = timeit.default_timer()
print('Time to fit:', end - start)
helpers.log_fit_time('WINE_NN', end - start)

# Predict quality with the trained best model
y_pred = best_model.predict(x_test)

helpers.produce_model_performance_summary(
    best_model,
    x_test,
    y_test,
    y_pred,
    grid_search=grid_search,
    output_location='wine_output/neural_net_summary.txt',
    cv=3,
    scoring='accuracy')

コード例 #4

                               'n_estimators':
                               [10, 20, 30, 40, 50, 60, 70, 80, 90, 100],
                           },
                           cv=3)

grid_search.fit(x_train, y_train)

print(grid_search.best_score_)
print(grid_search.best_params_)

# train the best model
best_model = grid_search.best_estimator_
# Time fitting best model
start = timeit.default_timer()
best_model.fit(x_train, y_train)
end = timeit.default_timer()
print('Time to fit:', end - start)
helpers.log_fit_time('WINE_BOOST', end - start)

# Predict quality with the trained best model
y_pred = best_model.predict(x_test)

helpers.produce_model_performance_summary(
    best_model,
    x_test,
    y_test,
    y_pred,
    output_location='wine_output/boost_summary.txt',
    cv=3,
    scoring='accuracy')

コード例 #5

helpers.log_fit_time('WINE_DT', end - start)

# Export decision tree to graphviz png
helpers.export_decision_tree_to_file(
    best_model,
    feature_names=feature_cols,
    class_names=['Low Quality', 'High Quality'],
    output_location=r'wine_output/decision_tree',
    format='png')

# Plot the learning curve vs train size after finding the best model
helpers.plot_learning_curve_vs_train_size(
    best_model,
    df,
    feature_cols,
    'quality_num',
    output_location='wine_output/best_model_num_samples_learning_curve.png')

# Predict income with the trained best model
y_pred = best_model.predict(x_test)

helpers.produce_model_performance_summary(
    best_model,
    x_test,
    y_test,
    y_pred,
    grid_search=grid_search,
    output_location='wine_output/decision_tree_summary.txt',
    cv=kfold,
    scoring='accuracy')