人工智能-机器学习-葵花8号卫星基于XGboost反演雾浓度与云顶高度

# coding: utf-8 # In[ ]: import numpy as np import scipy.io as scio import xgboost as xgb from xgboost.sklearn import XGBClassifier from xgboost.sklearn import XGBRegressor from sklearn import metrics from sklearn.model_selection import GridSearchCV,train_test_split from sklearn.externals import joblib np.set_printoptions(suppress= True) # In[ ]: # 加载数据 cloudy = scio.loadmat('data\data_cloud.mat')['cloudy'] clear = scio.loadmat('data\data_cloud.mat')['clear'] # cloudy和clear合并一个数组 data = np.concatenate((cloudy,clear),axis = 0) # 提取分类与特征 classification = data[:,2].copy() classification = np.where(classification > 0,1,0) features = np.delete(data,[2,3],axis=1) #交叉验证 x_train,x_test,y_train,y_test = train_test_split(features,classification,test_size = 0.25,random_state = 10) # In[ ]: # 迭代次数与learning rate param_test1 = {'n_estimators': list(range(100,201,50)),'learning_rate': np.linspace(0.05,0.3,6)} gsearch1 = GridSearchCV(estimator = XGBClassifier(max_depth= 5, # learning_rate= 0.1, # n_estimators= 150, n_jobs= -1, gamma= 0, reg_alpha= 0, reg_lambda= 1, seed= 10), param_grid = param_test1, scoring='accuracy',iid = False,n_jobs = -1,return_train_score = True) gsearch1.fit(x_train,y_train) gsearch1.cv_results_, gsearch1.best_score_, gsearch1.best_params_ # In[ ]: # max_depth param_test2 = {'max_depth': np.arange(5,11)} gsearch2 = GridSearchCV(estimator = XGBClassifier( # max_depth= 5, learning_rate= 0.3, n_estimators= 200, n_jobs= -1, gamma= 0, reg_alpha= 0, reg_lambda= 1, seed= 10), param_grid = param_test2, scoring='accuracy',iid = False,n_jobs = -1,return_train_score = True) gsearch2.fit(x_train,y_train) gsearch2.cv_results_, gsearch2.best_score_, gsearch2.best_params_ # In[ ]: # min_child_weight param_test3 = {'min_child_weight': np.arange(1,1002,200)} gsearch3 = GridSearchCV(estimator = XGBClassifier( max_depth= 10, learning_rate= 0.3, n_estimators= 200, n_jobs= -1, gamma= 0, reg_alpha= 0, reg_lambda= 1, # min_child_weight = 300, seed= 10), param_grid = param_test3, scoring='accuracy',iid = False,n_jobs = -1,return_train_score = True) gsearch3.fit(x_train,y_train) gsearch3.cv_results_, gsearch3.best_score_, gsearch3.best_params_ # In[ ]: # gamma param_test4 = {'gamma': np.linspace(0,0.5,5)} gsearch4 = GridSearchCV(estimator = XGBClassifier( max_depth= 10, learning_rate= 0.3, n_estimators= 200, n_jobs= -1, # gamma= 0, reg_alpha= 0, reg_lambda= 1, min_child_weight = 200, seed= 10), param_grid = param_test4, scoring='accuracy',iid = False,n_jobs = -1,return_train_score = True) gsearch4.fit(x_train,y_train) gsearch4.cv_results_, gsearch4.best_score_, gsearch4.best_params_ # In[ ]: # subsample param_test5 = {'subsample': [0.6,0.7,0.8,0.9,1.0]} gsearch5 = GridSearchCV(estimator = XGBClassifier( max_depth= 10, learning_rate= 0.3, n_estimators= 200, n_jobs= -1, gamma= 0.25, reg_alpha= 0, reg_lambda= 1, min_child_weight = 200, seed= 10), param_grid = param_test5, scoring='accuracy',iid = False,n_jobs = -1,return_train_score = True) gsearch5.fit(x_train,y_train) gsearch5.cv_results_, gsearch5.best_score_, gsearch5.best_params_ # In[ ]: # colsample_bytree param_test6 = {'colsample_bytree': [0.6,0.7,0.8,0.9,1.0]} gsearch6 = GridSearchCV(estimator = XGBClassifier( max_depth= 10, learning_rate= 0.3, n_estimators= 200, n_jobs= -1, gamma= 0.25, reg_alpha= 0, reg_lambda= 1, min_child_weight = 200, subsample = 1.0, seed= 10), param_grid = param_test6, scoring='accuracy',iid = False,n_jobs = -1,return_train_score = True) gsearch6.fit(x_train,y_train) gsearch6.cv_results_, gsearch6.best_score_, gsearch6.best_params_ # In[ ]: # 训练 xgb_skl = XGBClassifier(max_depth= 10, learning_rate= 0.1, n_estimators= 600, n_jobs= -1, gamma= 0.25, reg_alpha= 0, reg_lambda= 1, min_child_weight = 200, subsample = 1.0, colsample_bytree = 1.0, seed= 10) xgb_skl.fit(x_train,y_train) # 预测,测试集 y_pred = xgb_skl.predict(x_test) accuracy_test = metrics.accuracy_score(y_test, y_pred) print ("Accuracy(test) : %.4f" %accuracy_test) # 预测,训练集 y_pred = xgb_skl.predict(x_train) accuracy_train = metrics.accuracy_score(y_train, y_pred) print ("Accuracy(train) : %.4f" %accuracy_train) # In[ ]: # 微调训练 xgb_skl = XGBClassifier(max_depth= 13, learning_rate= 0.1, n_estimators= 150, n_jobs= -1, gamma= 0.11, reg_alpha= 0, reg_lambda= 3, min_child_weight = 1, subsample = 1, colsample_bytree = 0.9, seed= 10) xgb_skl.fit(x_train,y_train) # 预测,测试集 y_pred = xgb_skl.predict(x_test) accuracy_test = metrics.accuracy_score(y_test, y_pred) print ("Accuracy(test) : %.4f" %accuracy_test) # 预测,训练集 y_pred = xgb_skl.predict(x_train) accuracy_train = metrics.accuracy_score(y_t