#对大于97.5%分位数的因子值，或小于2.5%分位数的因子值进行调整  容易理解，去除两头，只需调节参数def extreme_percentile(data,min=0.025,max=0.975):    p = data.quantile([min,max])                  # 得到上下限的值return data.clip(p.iloc[0],p.iloc[1])  # 超出上下限的值，赋值为上下限

from sklearn.preprocessing import OneHotEncoder# 构造数据X_train = [['male', 'from US', 'uses Safari'],      ['female', 'from Europe', 'uses Firefox'],      ['female', 'from China', 'uses Safari']]  # 编码器encoder = OneHotEncoder()encoder = encoder.fit(X_train)# 编码X = [['female', 'from Europe', 'uses Safari']]X_transform = encoder.transform(X) X_transform.toarray()     # 默认返回的是稀疏矩阵, 用toarray()方法可以转为np.array格式>> array([[1., 0., 0., 1., 0., 0., 1.]])

import pandas as pdimport numpy as npfrom sklearn.ensemble import RandomForestClassifier  # 随机森林分类from sklearn.model_selection import train_test_split,GridSearchCV,cross_val_scorefrom sklearn.metrics import accuracy_score # 引入准确度评分函数from sklearn.metrics import mean_squared_errorfrom sklearn.metrics import confusion_matrix, accuracy_score,classification_reportfrom sklearn import preprocessingimport warningswarnings.filterwarnings('ignore')import matplotlib.pyplot as pltimport matplotlibmatplotlib.rc("font", family='Microsoft YaHei')

price_classify_data = pd.read_csv('yichang_deal.csv')X = price_classify_data[['zone_label','street_label','type_label','toward_label','area','floor_label','decorate_label','structure_label','follow']]Y = price_classify_data['price_label']# 标准化X = preprocessing.scale(X)X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.3, random_state=200)  # 随机数种子

###调n_estimators参数ScoreAll = []for i in range(90,200,10):    DT = RandomForestClassifier(n_estimators = i,random_state = 1) #,criterion = 'entropy'    score = cross_val_score(DT,X_train,Y_train,cv=6).mean()    ScoreAll.append([i,score])ScoreAll = np.array(ScoreAll)
max_score = np.where(ScoreAll==np.max(ScoreAll[:,1]))[0][0] ##这句话看似很长的，其实就是找出最高得分对应的索引print("最优参数以及最高得分:",ScoreAll[max_score])  plt.figure(figsize=[20,5])plt.plot(ScoreAll[:,0],ScoreAll[:,1],'r-o',label='最高得分',color='orange')plt.xlabel('n_estimators参数')plt.ylabel('分数')plt.grid()plt.legend()plt.show()# [130.           0.77857096]

# 探索max_depth的最佳参数ScoreAll = []  for i in range(10,30,3):      DT = RandomForestClassifier(n_estimators = 130,random_state = 1,max_depth =i ) #,criterion = 'entropy'      score = cross_val_score(DT,X_train,Y_train,cv=10).mean()      ScoreAll.append([i,score])  ScoreAll = np.array(ScoreAll)  
max_score = np.where(ScoreAll==np.max(ScoreAll[:,1]))[0][0]print("最优参数以及最高得分:",ScoreAll[max_score])    plt.figure(figsize=[20,5])  plt.plot(ScoreAll[:,0],ScoreAll[:,1])plt.xlabel('max_depth最佳参数')plt.ylabel('分数')plt.grid()plt.legend()plt.show()

model = RandomForestClassifier(n_estimators=128,max_depth=22,random_state=1) # min_samples_leaf=11model.fit(X_train,Y_train)score = model.score(X_test,Y_test)print('模型分数:',score)# 计算在训练集和测试集上的预测均方根误差model_lab = model.predict(X_train)model_pre = model.predict(X_test)train_mse = mean_squared_error(Y_train,model_lab)test_mse = mean_squared_error(Y_test,model_pre)print('训练数据集上的均方根误差：',train_mse)print('测试数据集上的均方根误差：',test_mse)# 模型使用与评估y_pred = model.predict(X_test)print("准确率: %.3f" % accuracy_score(Y_test, y_pred))

from sklearn.metrics import confusion_matriximport matplotlib.ticker as ticker
cm = confusion_matrix(Y_test, y_pred,labels=[0,1,-1])print('混淆矩阵：\n', cm)labels=['-1','0','1']from sklearn.metrics import ConfusionMatrixDisplaycm_display = ConfusionMatrixDisplay(cm,display_labels=labels).plot()# 混淆矩阵热力图def cm_plot(cm,labels):# 设置字体、字号    plt.rc('font',family='Times New Roman',size='14')
    ind_array = np.arange(len(labels))    x, y = np.meshgrid(ind_array, ind_array)
for x_val, y_val in zip(x.flatten(), y.flatten()):# 转换为百分数，需同时更改colorbar标签(当然也可以直接以样本数表示，后续colorbar无需变动)        c = round(cm[y_val][x_val]/sum(cm[y_val])*100,1)        plt.text(x_val, y_val, c, color='b', va='center', ha='center')
    plt.imshow(cm, interpolation='nearest', cmap=plt.cm.binary)
# colorbar设置    cb = plt.colorbar()    tick_locator = ticker.MaxNLocator(nbins=5)    cb.locator = tick_locator
    x_label = np.array(range(len(labels)))    plt.xticks(x_label, labels)    plt.yticks(x_label, labels, rotation=90)    plt.ylabel('Actual label')    plt.xlabel('Predict label')
    ax = plt.gca()    tick_marks = np.array(range(len(labels))) + 0.5    ax.set_xticks(tick_marks, minor=True)    ax.set_yticks(tick_marks, minor=True)    ax.xaxis.set_ticks_position('none')    ax.yaxis.set_ticks_position('none')    ax.grid(True, which='minor', linestyle='-')
    plt.tight_layout()    plt.show()
cm_plot(cm,labels)

features = X.columnsimportances = model.feature_importances_df = pd.DataFrame()df['特征'] = featuresdf['重要性'] = importancesdf = df.sort_values('重要性',ascending=False)# df.iloc[2,1] = df.iloc[2,1]+0.2xx = df['特征']yy = df['重要性']colors = ['darkgreen','gold','red','navy','orange','lime','plum','sienna']plt.figure(figsize=(15,5))plt.barh(xx,yy,0.4,color=colors,alpha=0.8)#color参数传入颜色列表，可以在一幅图中显示不同颜色plt.xlabel('Importance score')plt.ylabel('variable')plt.title('RandomForestRegressor')plt.grid()plt.show()