def concrete():
print("Started training linear regressor on concrete data set.")
x_train, x_test, y_train, y_test = data_processing.concrete()
result = {}
# LR
lr_best_model, lr_params = regression_cv.linear_regression(x_train, y_train, fold=3, iterations=20)
result[constant.LR] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"concrete", "Linear Regression")
# decision tree
lr_best_model, lr_params = regression_cv.decision_tree(x_train, y_train, max_depth=20, fold=3, iterations=20)
result[constant.DECISION_TREE] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"concrete", "Decision tree")
# Random Forest
lr_best_model, lr_params = regression_cv.random_forest(x_train, y_train, max_estimator=100, fold=3, iterations=20)
result[constant.RANDOM_FOREST] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"concrete", "Random Forest")
# SVM
lr_best_model, lr_params = regression_cv.SVR(x_train, y_train,
# ['linear', 'poly', 'rbf', 'sigmoid'],
['linear'],
0.01, 100, 1, 1000,
fold=3, iterations=10)
result[constant.SVC] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"concrete", "SVM")
# AdaBoost
lr_best_model, lr_params = regression_cv.ada_boost_regression(x_train, y_train, no_estimators=50,
fold=3, iterations=20)
result[constant.ADABOOST] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"concrete", "AdaBoost")
# NeuralNet (not finish)
#lr_best_model, lr_params = regression_cv.NeuralNetworkRegression(x_train, y_train,
# hidden_layer_sizes=[(100,), (10,)],
# alphas=[0.01, 0.05, 0.5, 1],
# max_iter=[10, 20, 50, 100],
# fold=3, iterations=20)
#result[constant.NN] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
# "concrete", "NeuralNet")
#GaussianProcessRegressor
lr_best_model, lr_params = regression_cv.GPRegressor(x_train, y_train, fold=3, iterations=10)
result[constant.GP] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"concrete", "Gaussian process")
export_result(result, "result/concrete.json")
def red_wine_quality():
print("Started training linear regressor on red wine quality data set.")
x_train, x_test, y_train, y_test = data_processing.red_wine_quality()
result = {}
# LR
lr_best_model, lr_params = regression_cv.linear_regression(x_train, y_train, fold=3, iterations=20)
result[constant.LR] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"Wine quality", "Linear Regression")
# decision tree
lr_best_model, lr_params = regression_cv.decision_tree(x_train, y_train, max_depth=20, fold=3, iterations=20)
result[constant.DECISION_TREE] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"Wine quality", "Decision tree")
# Random Forest
lr_best_model, lr_params = regression_cv.random_forest(x_train, y_train, max_estimator=100, fold=3, iterations=20)
result[constant.RANDOM_FOREST] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"Wine quality", "Random Forest")
# SVM
lr_best_model, lr_params = regression_cv.SVR(x_train, y_train,
['linear', 'poly', 'rbf', 'sigmoid'],
0.01, 100, 1, 1000,
fold=3, iterations=20)
result[constant.SVC] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"Wine quality", "SVM")
# AdaBoost
lr_best_model, lr_params = regression_cv.ada_boost_regression(x_train, y_train, no_estimators=50,
fold=3, iterations=20)
result[constant.ADABOOST] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"Wine quality", "AdaBoost")
# NeuralNet
lr_best_model, lr_params = regression_cv.NeuralNetworkRegression(x_train, y_train,
hidden_layer_sizes=[(), (10,)],
alphas=[0.01, 0.05, 0.5, 1],
max_iter=[100, 200, 500, 1000],
fold=3, iterations=20)
result[constant.NN] = evaluate_regressor(x_train, x_test, y_train, y_test, lr_best_model, lr_params,
"Wine quality", "NeuralNet")
export_result(result, "result/red_wine.json")