사용 방법
KNeighborsRegressor - man
from sklearn.neighbors import KNeighborsRegressor
neigh = KNeighborsRegressor()
param_grid4 = { 'n_neighbors': np.arange(1,50,2)}
grid4 = GridSearchCV(neigh, param_grid = param_grid4,
cv=RepeatedKFold(n_splits=5, n_repeats=5, random_state = 123),
return_train_score=True)
grid4.fit(x_man_train,y_man_train)
print(f"best parameters: {grid4.best_params_}")
print(f"valid-set score: {grid4.score(x_man_valid, y_man_valid):.3f}")best parameters: {'n_neighbors': 9}
valid-set score: 0.622
neigh = KNeighborsRegressor(n_neighbors = 9)
y_woman_pred = cross_val_predict(neigh, x_man_train, y_man_train, cv=5)
plt.plot([15, 85], [15, 85], color='k')
plt.scatter(y_woman_pred, y_man_train, alpha=.5, s=4)
KNeighborsRegressor - woman
neigh = KNeighborsRegressor()
param_grid4 = { 'n_neighbors': np.arange(1,50,2)}
grid4 = GridSearchCV(neigh, param_grid = param_grid4,
cv=RepeatedKFold(n_splits=5, n_repeats=5, random_state = 123),
return_train_score=True)
grid4.fit(x_woman_train,y_woman_train)
print(f"best parameters: {grid4.best_params_}")
print(f"valid-set score: {grid4.score(x_woman_valid, y_woman_valid):.3f}")best parameters: {'n_neighbors': 7}
valid-set score: 0.623
neigh = KNeighborsRegressor(n_neighbors=7)
y_woman_pred = cross_val_predict(neigh, x_woman_train, y_woman_train, cv=5)
plt.plot([15, 85], [15, 85], color='k')
plt.scatter(y_woman_pred, y_woman_train, alpha=.5, s=4)
SGDRegressor - man
from sklearn.linear_model import SGDRegressor
sgd = SGDRegressor(max_iter=1000, tol=1e-3)
param_grid5 = {'alpha' : np.arange(0.1,10,0.1), 'l1_ratio' : np.arange(0.1,1,0.1)}
grid5 = GridSearchCV(sgd, param_grid = param_grid5,
cv = RepeatedKFold(n_splits=5, n_repeats=3, random_state = 123),
return_train_score=True)
grid5.fit(x_man_train,y_man_train)
print(f"best parameters: {grid5.best_params_}")
print(f"valid-set score: {grid5.score(x_man_valid, y_man_valid):.3f}")best parameters: {'alpha': 0.6, 'l1_ratio': 0.4}
valid-set score: 0.751
results = pd.DataFrame(grid5.cv_results_)
results.plot('param_alpha', 'mean_train_score')
results.plot('param_alpha', 'mean_test_score', ax=plt.gca())
plt.fill_between(results.param_alpha.astype(np.float64),
results['mean_train_score'] + results['std_train_score'],
results['mean_train_score'] - results['std_train_score'], alpha=0.2)
plt.fill_between(results.param_alpha.astype(np.float64),
results['mean_test_score'] + results['std_test_score'],
results['mean_test_score'] - results['std_test_score'], alpha=0.2)
plt.legend()
sgd = SGDRegressor(max_iter=1000, tol=1e-3, alpha = 0.6, l1_ratio = 0.4)
y_man_pred = cross_val_predict(sgd, x_man_train, y_man_train, cv=5)
plt.plot([15, 85], [15, 85], color='k')
plt.scatter(y_man_pred, y_man_train, alpha=.5, s=4)
SGDRegressor - woman
sgd = SGDRegressor(max_iter=1000, tol=1e-3)
param_grid5 = {'alpha' : np.arange(0.1,10,0.1), 'l1_ratio' : np.arange(0.1,1,0.1)}
grid5 = GridSearchCV(sgd, param_grid = param_grid5,
cv = RepeatedKFold(n_splits=5, n_repeats=3, random_state = 123),
return_train_score=True)
grid5.fit(x_woman_train,y_woman_train)
print(f"best parameters: {grid5.best_params_}")
print(f"valid-set score: {grid5.score(x_woman_valid, y_woman_valid):.3f}")best parameters: {'alpha': 0.30000000000000004, 'l1_ratio': 0.2}
valid-set score: 0.552
results = pd.DataFrame(grid5.cv_results_)
results.plot('param_alpha', 'mean_train_score')
results.plot('param_alpha', 'mean_test_score', ax=plt.gca())
plt.fill_between(results.param_alpha.astype(np.float64),
results['mean_train_score'] + results['std_train_score'],
results['mean_train_score'] - results['std_train_score'], alpha=0.2)
plt.fill_between(results.param_alpha.astype(np.float64),
results['mean_test_score'] + results['std_test_score'],
results['mean_test_score'] - results['std_test_score'], alpha=0.2)
plt.legend()
sgd = SGDRegressor(max_iter=1000, tol=1e-3, alpha = 0.3, l1_ratio = 0.2)
y_woman_pred = cross_val_predict(sgd, x_woman_train, y_woman_train, cv=5)
plt.plot([15, 85], [15, 85], color='k')
plt.scatter(y_woman_pred, y_woman_train, alpha=.5, s=4)
TweedieRegressor - man
from sklearn.linear_model import TweedieRegressor
reg = TweedieRegressor(power=1, link='log')
param_grid5 = {'alpha' : np.arange(10,20,0.1)}
grid5 = GridSearchCV(reg, param_grid = param_grid5,
cv = RepeatedKFold(n_splits=5, n_repeats=3, random_state = 123),
return_train_score=True)
grid5.fit(x_man_train,y_man_train)
print(f"best parameters: {grid5.best_params_}")
print(f"valid-set score: {grid5.score(x_man_valid, y_man_valid):.3f}")best parameters: {'alpha': 19.899999999999963}
valid-set score: 0.712
results = pd.DataFrame(grid5.cv_results_)
results.plot('param_alpha', 'mean_train_score')
results.plot('param_alpha', 'mean_test_score', ax=plt.gca())
plt.fill_between(results.param_alpha.astype(np.float64),
results['mean_train_score'] + results['std_train_score'],
results['mean_train_score'] - results['std_train_score'], alpha=0.2)
plt.fill_between(results.param_alpha.astype(np.float64),
results['mean_test_score'] + results['std_test_score'],
results['mean_test_score'] - results['std_test_score'], alpha=0.2)
plt.legend()
reg = TweedieRegressor(power=1, link='log', alpha = 19.9)
y_man_pred = cross_val_predict(reg, x_man_train, y_man_train, cv=5)
plt.plot([15, 85], [15, 85], color='k')
plt.scatter(y_man_pred, y_man_train, alpha=.5, s=4)
TweedieRegressor - woman
reg = TweedieRegressor(power=1, link='log')
param_grid5 = {'alpha' : np.arange(10,20,0.1)}
grid5 = GridSearchCV(reg, param_grid = param_grid5,
cv = RepeatedKFold(n_splits=5, n_repeats=3, random_state = 123),
return_train_score=True)
grid5.fit(x_woman_train,y_woman_train)
print(f"best parameters: {grid5.best_params_}")
print(f"valid-set score: {grid5.score(x_woman_valid, y_woman_valid):.3f}")best parameters: {'alpha': 13.399999999999988}
valid-set score: 0.514
results = pd.DataFrame(grid5.cv_results_)
results.plot('param_alpha', 'mean_train_score')
results.plot('param_alpha', 'mean_test_score', ax=plt.gca())
plt.fill_between(results.param_alpha.astype(np.float64),
results['mean_train_score'] + results['std_train_score'],
results['mean_train_score'] - results['std_train_score'], alpha=0.2)
plt.fill_between(results.param_alpha.astype(np.float64),
results['mean_test_score'] + results['std_test_score'],
results['mean_test_score'] - results['std_test_score'], alpha=0.2)
plt.legend()
reg = TweedieRegressor(power=1, link='log', alpha = 13.4)
y_woman_pred = cross_val_predict(reg, x_woman_train, y_woman_train, cv=5)
plt.plot([15, 85], [15, 85], color='k')
plt.scatter(y_woman_pred, y_woman_train, alpha=.5, s=4)
DecisionTreeRegressor - man
from sklearn import tree
clf = tree.DecisionTreeRegressor()
param_grid5 = {'max_depth' : np.arange(1,20,1)}
grid5 = GridSearchCV(clf, param_grid = param_grid5, cv = RepeatedKFold(n_splits=5, n_repeats=3), return_train_score=True)
grid5.fit(x_man_train,y_man_train)
print(f"best parameters: {grid5.best_params_}")
print(f"valid-set score: {grid5.score(x_man_valid, y_man_valid):.3f}")best parameters: {'max_depth': 1}
valid-set score: 0.505
DecisionTreeRegressor - woman
from sklearn import tree
clf = tree.DecisionTreeRegressor()
param_grid5 = {'max_depth' : np.arange(1,20,1)}
grid5 = GridSearchCV(clf, param_grid = param_grid5, cv = RepeatedKFold(n_splits=5, n_repeats=3), return_train_score=True)
grid5.fit(x_woman_train,y_woman_train)
print(f"best parameters: {grid5.best_params_}")
print(f"valid-set score: {grid5.score(x_woman_valid, y_woman_valid):.3f}")best parameters: {'max_depth': 2}
valid-set score: 0.609
GradientBoostingRegressor - man
from sklearn.ensemble import GradientBoostingRegressor
reg = GradientBoostingRegressor(random_state=0)
param_grid5 = {'n_estimators' : np.arange(60,80,1)}
grid5 = GridSearchCV(reg, param_grid = param_grid5,
cv = RepeatedKFold(n_splits=5, n_repeats=3, random_state = 123),
return_train_score=True)
grid5.fit(x_man_train,y_man_train)
print(f"best parameters: {grid5.best_params_}")
print(f"valid-set score: {grid5.score(x_man_valid, y_man_valid):.3f}")best parameters: {'n_estimators': 69}
valid-set score: 0.741
reg = GradientBoostingRegressor(random_state=0, n_estimators = 69)
y_man_pred = cross_val_predict(reg, x_man_train, y_man_train, cv=5)
plt.plot([15, 85], [15, 85], color='k')
plt.scatter(y_man_pred, y_man_train, alpha=.5, s=4)
GradientBoostingRegressor - woman
reg = GradientBoostingRegressor(random_state=0)
param_grid5 = {'n_estimators' : np.arange(80,90,1)}
grid5 = GridSearchCV(reg, param_grid = param_grid5,
cv = RepeatedKFold(n_splits=5, n_repeats=3, random_state = 123),
return_train_score=True)
grid5.fit(x_woman_train,y_woman_train)
print(f"best parameters: {grid5.best_params_}")
print(f"valid-set score: {grid5.score(x_woman_valid, y_woman_valid):.3f}")best parameters: {'n_estimators': 86}
valid-set score: 0.676
reg = GradientBoostingRegressor(random_state=0, n_estimators = 81)
y_woman_pred = cross_val_predict(reg, x_woman_train, y_woman_train, cv=5)
plt.plot([15, 85], [15, 85], color='k')
plt.scatter(y_woman_pred, y_woman_train, alpha=.5, s=4)
KHU, SWCON