def runLarsRegressor(self):
lm = Lars(fit_intercept=True, normalize=True)
print("Lars Regressor\n")
lm.fit(self.m_X_train, self.m_y_train)
predictY = lm.predict(self.m_X_test)
score = lm.score(self.m_X_test, self.m_y_test)
predictTraingY = lm.predict(self.m_X_train)
self.displayPredictPlot(predictY)
self.displayResidualPlot(predictY, predictTraingY)
self.dispalyModelResult(lm, predictY, score)
#LarsCV: fit_intercept, verbose, normalize, cv from sklearn.linear_model import LarsCV, Lars from sklearn.datasets import make_regression import matplotlib.pyplot as plt X, y = make_regression(n_samples=200, noise=4.0, random_state=0) reg = LarsCV(cv=5).fit(X, y) reg.score(X, y) reg.alpha_ pred = reg.predict(X[:, ]) plt.scatter(X[:, 0], y, color='black') plt.scatter(X[:, 0], pred, color='red') plt.show() reg2 = Lars().fit(X, y) reg2.score(X, y) reg2.alpha_ pred = reg2.predict(X[:, ]) #%% LassoLars: alpha, fit_intercept, normalize #LassoLarsCV: alpha, fit_intercept, normalize, cv from sklearn import linear_model reg = linear_model.LassoLars(alpha=0.01) reg.fit([[-1, 1], [0, 0], [1, 1]], [-1, 0, -1]) print(reg.coef_) reg2 = linear_model.LassoLarsCV() reg2.fit([[-1, 1], [0, 0], [1, 1]], [-1, 0, -1])
# LARS Regression import numpy as np from sklearn import datasets from sklearn.linear_model import Lars # load the diabetes datasets dataset = datasets.load_diabetes() # fit a LARS model to the data model = Lars() model.fit(dataset.data, dataset.target) print(model) # make predictions expected = dataset.target predicted = model.predict(dataset.data) # summarize the fit of the model mse = np.mean((predicted-expected)**2) print(mse) print(model.score(dataset.data, dataset.target))
# LARS Regression # The Least Angle Regression (LARS) method is a computationally efficient algorithm for fitting a regression model. # It is useful for highdimensional data and is commonly used in conjunction with regularization (such as LASSO). import numpy as np from sklearn import datasets from sklearn.linear_model import Lars # load the diabetes datasets dataset = datasets.load_diabetes() # fit a LARS model to the data model = Lars() model.fit(dataset.data, dataset.target) print(model) # make predictions expected = dataset.target predicted = model.predict(dataset.data) # summarize the fit of the model mse = np.mean((predicted-expected)**2) print(mse) print(model.score(dataset.data, dataset.target))
# PCA + Elastic Net elasticnet = ElasticNet(l1_ratio=0.5) elasticnet.fit(reduced_training_features, training_labels) preds = elasticnet.predict(reduced_testing_features) score = elasticnet.score(reduced_testing_features,testing_labels) print 'PCA + ElasticNet Results:' print 'R2 score:', score print 'MAE:', mean_absolute_error(testing_labels,preds) # Least-Angle Regression (LARS) from sklearn.linear_model import Lars lars = Lars() lars.fit(training_features, training_labels) preds = lars.predict(testing_features) score = lars.score(testing_features,testing_labels) print 'LARS Results:' print 'R2 score:', score print 'MAE:', mean_absolute_error(testing_labels,preds), '\n' # PCA + LARS lars = Lars() lars.fit(reduced_training_features, training_labels) preds = lars.predict(reduced_testing_features) score = lars.score(reduced_testing_features,testing_labels) print 'PCA + LARS Results:' print 'R2 score:', score print 'MAE:', mean_absolute_error(testing_labels,preds) # Orthogonal Matching Pursuit from sklearn.linear_model import OrthogonalMatchingPursuit
可以快速改造成lasso
缺点：
因为模型是对残差进行迭代设计，所以对噪声敏感
'''
rg = Lars(fit_intercept=True,
verbose=False,
normalize=True,
precompute='auto',
n_nonzero_coefs=500,
eps=2.2204460492503131e-16,
copy_X=True,
fit_path=True,
positive=False)
rg.fit(X_train, Y_train)
Y_pre = rg.predict(X_test)
rg.score(X_test, Y_test)
rg.coef_
rg.intercept_
'''
fit_intercept 是否训练截距
verbose 冗长度
normalize 归一化否
precompute 是否使用Gram矩阵来加速
n_nonzero_coefs 非零系数的目标数
eps 精确度，计算某个值时用到
copy_X 是否覆盖模型中的X
fit_path 不太理解，暂时应该也用不到
positive 设置强制系数为正的嘛？
'''
predictions = linreg.predict(x_test)
# Plot predictions and y_test
plt.figure()
plt.plot(predictions, label='Predictions')
plt.plot(pd.Series(predictions).rolling(5).mean(),
label='rolling predictions')
plt.plot(y_test.values,
label='Shifted Currencies ( y_test values',
color='grey')
plt.plot(cu.loc[test_index, currency].values, label='UNSHIFTED')
plt.legend()
plt.show()
# Print Score and summary
score = linreg.score(x_test, y_test)
print('SCORE: {:.2f}'.format(score))
print('Explained Variance: {}'\
.format(explained_variance_score(y_test, predictions)))
###############################################################################
# Use two currency predictions to predict instrument
###############################################################################
if 0:
# Get Parameters
if False:
cur1 = predictions
cur2 = predictions
# Choose Instrument
from sklearn.linear_model import Lars
from sklearn.model_selection import train_test_split
lars = Lars(fit_intercept=False,
normalize=False,
n_nonzero_coefs=100,
verbose=True)
x_train, x_test, y_train, y_test = train_test_split(x_scaled,
y_scaled,
test_size=0.2,
random_state=42)
lars.fit(x_train, y_train)
#print(x_test[0])
lars.predict(x_test)[0]
lars.score(x_test, y_test)
lars.get_params()
beta = generate_random_points(n=10, p=10)
scaler.fit(beta)
scaler.fit_transform(beta)
beta = scaler.fit_transform(beta)
epsilons = generate_random_points(n=100, p=10)
#print(epsilons)
y = [[] for _ in range(10)]
for k in range(10):
y[k] = np.matmul(beta, np.asarray(x[k])) + epsilons[k]
x_scaled = scale(x)