頁共31頁附錄A程序代碼importpandasaspdimportseabornassnsfromsklearn.treeimportDecisionTreeClassifier,plot_treeimportmatplotlib.pyplotaspltfromsklearn.model_selectionimporttrain_test_splitfromsklearn.treeimportDecisionTreeClassifierfromsklearn.metricsimportclassification_reportfromsklearn.model_selectionimportcross_val_scorefromsklearn.metricsimportf1_scorefromsklearn.metricsimportaccuracy_scorefromsklearn.metricsimportconfusion_matriximportwarningswarnings.filterwarnings('ignore')data=pd.read_csv('train.csv')#數(shù)據(jù)集print(data.T)#數(shù)據(jù)集倒置print(data.isna().sum())#檢查缺失值print(data.describe())#統(tǒng)計信息（非空，平均，標(biāo)準(zhǔn)差，最小，下四分位，中位，上四分位，最大）print(())#訓(xùn)練集的基本信息plt.figure(figsize=(20,12))sns.heatmap(data.corr(),annot=True)plt.show()#數(shù)值之間相關(guān)熱性圖x=data.drop(['price_range'],axis=1)#特征集y=data.price_range.values.reshape(-1,1)#目標(biāo)變量集x_train,x_test,y_train,y_test=train_test_split(x,y,test_size=0.2,random_state=0)#劃分訓(xùn)練集，測試集#嘗試不同的深度參數(shù)depths=[1,2,3,4,5,6,7,8,9,10,None]#None表示不限制深度best_accuracy=0best_depth=Nonefordepthindepths:clf=DecisionTreeClassifier(max_depth=depth,random_state=42)clf.fit(x_train,y_train)y_pred=clf.predict(x_test)#預(yù)測測試集accuracy=accuracy_score(y_test,y_pred)#計算準(zhǔn)確率print(f"Depth:{depth},Accuracy:{accuracy}")#更新最佳準(zhǔn)確率和最佳深度ifaccuracy>best_accuracy:best_accuracy=accuracybest_depth=depthprint(f"Bestdepthfound:{best_depth}withaccuracy:{best_accuracy}")model=DecisionTreeClassifier(max_depth=8)#決策樹分類器，深度為8model.fit(x_train,y_train)#訓(xùn)練集訓(xùn)練決策樹分類器y_pred_train=model.predict(x_train)accuracy_train=accuracy_score(y_train,y_pred_train)print(f'Accuracyontrainingset:{accuracy_train:.2f}')#決策樹模型訓(xùn)練集中的準(zhǔn)確率y_pred=model.predict(x_test)print("_________________________________")print(accuracy_score(y_test,y_pred))#測試集的準(zhǔn)確率print("________________________________")print(classification_report(y_test,y_pred))#模型性能的主要分類指標(biāo)（精確度（Precision）召回率（Recall）F1分?jǐn)?shù)（F1-score）支持?jǐn)?shù)（Support）print("_____________________________________")print(confusion_matrix(y_test,y_pred))#每個類別的精確度（precision）、召回率（recall）、F1分?jǐn)?shù)以及支持?jǐn)?shù)（即該類別的樣本數(shù)）#繪制混淆矩陣熱圖cm=confusion_matrix(y_test,y_pred)plt.figure(figsize=(10,7))sns.heatmap(cm,annot=True,cmap='coolwarm',fmt='d')#使用'coolwarm'色彩映射，并設(shè)置fmt='d'來顯示整數(shù)plt.xlabel('PredictedLabel')#X軸標(biāo)簽plt.ylabel('TrueLabel')#Y軸標(biāo)簽plt.title('ConfusionMatrix')#添加標(biāo)題plt.show()#顯示熱圖f1_macro=f1_score(y_test,y_pred,average='macro')#'micro'計算全局的F1分?jǐn)?shù)（所有類別的TP,FP,FN之和）f1_micro=f1_score(y_test,y_pred,average='micro')#'macro'計算每個類別的F1分?jǐn)?shù)，然后取平均值（不考慮類別樣本數(shù)量）f1_weighted=f1_score(y_test,y_pred,average='weighted')#'weighted'計算每個類別的F1分?jǐn)?shù)，然后取平均值，權(quán)重是各類別樣本數(shù)量的比例print('MacroF1Score:',f1_macro)print('MicroF1Score:',f1_micro)print('WeightedF1Score:',f1_weighted)#通過限制樹的最大深度來剪枝clf=DecisionTreeClassifier(max_depth=8,random_state=42)#訓(xùn)練模型clf.fit(x_train,y_train)#將類標(biāo)簽從整數(shù)轉(zhuǎn)換為字符串class_names=[str(cls)forclsinmodel.classes_]fig,ax=plt.subplots(figsize=(10,5))#調(diào)整畫布大小以適應(yīng)樹的大小plot_tree(model,feature_names=x_train.columns,#特征名稱class_names=class_names,#類標(biāo)簽名稱，轉(zhuǎn)換為字符串列表filled=True,#是否填充顏色來表示