機(jī)器學(xué)習(xí)最強(qiáng)調(diào)參方法！高斯過(guò)程與貝葉斯優(yōu)化

import numpy as npfrom sklearn.gaussian_process import GaussianProcessRegressorfrom sklearn.gaussian_process.kernels import ConstantKernel, RBF
# 定義目標(biāo)函數(shù)def target(x):    return np.exp(-(x - 2)**2) + np.exp(-(x - 6)**2/10) + 1/ (x**2 + 1)
# 訓(xùn)練和測(cè)試數(shù)據(jù)X_train = np.array([-1, 2, 4, 5, 7]).reshape(-1, 1)y_train = target(X_train)X_test = np.linspace(-2, 10, 10000).reshape(-1, 1)
# 高斯過(guò)程擬合kernel = ConstantKernel(constant_value=0.2,constant_value_bounds=(1e-4, 1e4)) *                         RBF(length_scale=0.5, length_scale_bounds=(1e-4, 1e4))gpr = GaussianProcessRegressor(kernel=kernel, n_restarts_optimizer=2)gpr.fit(X_train, y_train)mu, cov = gpr.predict(X_test, return_cov=True)y_test = mu.ravel()uncertainty = 1.96 * np.sqrt(np.diag(cov))
# 繪圖plt.figure()plt.title("l=%.1f sigma_f=%.1f" % (gpr.kernel_.k2.length_scale, gpr.kernel_.k1.constant_value))plt.fill_between(X_test.ravel(), y_test + uncertainty, y_test - uncertainty, alpha=0.1)plt.plot(X_test, y_test, label="predict")plt.scatter(X_train, y_train, label="train", c="red", marker="D")plt.legend()

pip install bayesian-optimization

from bayes_opt import BayesianOptimization from bayes_opt import UtilityFunctionx = np.linspace(-2, 10, 10000).reshape(-1, 1)y = target(x)# 創(chuàng)建貝葉斯優(yōu)化器optimizer = BayesianOptimization(target, {'x': (-2, 10)}, random_state=27)# 隨機(jī)初始化兩個(gè)點(diǎn)optimizer.maximize(init_points=2, n_iter=0, kappa=5)
# 定義后驗(yàn)計(jì)算函數(shù)def posterior(optimizer, x_obs, y_obs, grid):    optimizer._gp.fit(x_obs, y_obs)    mu, sigma = optimizer._gp.predict(grid, return_std=True)    return mu, sigma

import pandas as pdfrom sklearn.model_selection import train_test_splitdata = pd.read_csv("flights.csv")data = data.sample(frac=0.01, random_state=10)data = data[["MONTH","DAY","DAY_OF_WEEK","AIRLINE","FLIGHT_NUMBER","DESTINATION_AIRPORT",  "ORIGIN_AIRPORT","AIR_TIME", "DEPARTURE_TIME","DISTANCE","ARRIVAL_DELAY"]]
data["ARRIVAL_DELAY"] = (data["ARRIVAL_DELAY"]>10)*1cols = ["AIRLINE","FLIGHT_NUMBER","DESTINATION_AIRPORT","ORIGIN_AIRPORT"]for item in cols:    data[item] = data[item].astype("category").cat.codes +1
X_train, X_test, y_train, y_test = train_test_split(data.drop(["ARRIVAL_DELAY"],           axis=1), data["ARRIVAL_DELAY"], random_state=10, test_size=0.3)print(X_train.shape, y_train.shape, X_test.shape, y_test.shape)

import xgboost as xgbfrom bayes_opt import BayesianOptimization
# 定義目標(biāo)優(yōu)化函數(shù)def xgb_evaluate(min_child_weight,                 colsample_bytree,                 max_depth,                 subsample,                 gamma,                 alpha):
    params['min_child_weight'] = int(min_child_weight)    params['cosample_bytree'] = max(min(colsample_bytree, 1), 0)    params['max_depth'] = int(max_depth)    params['subsample'] = max(min(subsample, 1), 0)    params['gamma'] = max(gamma, 0)    params['alpha'] = max(alpha, 0)
    cv_result = xgb.cv(params, dtrain, num_boost_round=num_rounds, nfold=5,             seed=random_state,             callbacks=[xgb.callback.early_stop(50)])
    return cv_result['test-auc-mean'].values[-1]
# 定義相關(guān)參數(shù)num_rounds = 3000random_state = 2021num_iter = 25init_points = 5params = {    'eta': 0.1,    'silent': 1,    'eval_metric': 'auc',    'verbose_eval': True,    'seed': random_state}
# 創(chuàng)建貝葉斯優(yōu)化實(shí)例# 并設(shè)定參數(shù)搜索范圍xgbBO = BayesianOptimization(xgb_evaluate, {'min_child_weight': (1, 20),                                            'colsample_bytree': (0.1, 1),                                            'max_depth': (5, 15),                                            'subsample': (0.5, 1),                                            'gamma': (0, 10),                                            'alpha': (0, 10),                                            })
# 執(zhí)行調(diào)優(yōu)過(guò)程xgbBO.maximize(init_points=init_points, n_iter=num_iter)