Visualizers
Visualizers也是estimators（從數(shù)據(jù)中習得的對象），其主要任務是產(chǎn)生可對模型選擇過程有更深入了解的視圖。從Scikit-Learn來看，當可視化數(shù)據(jù)空間或者封裝一個模型estimator時，其和轉(zhuǎn)換器（transformers）相似，就像"ModelCV" (比如 RidgeCV, LassoCV )的工作原理一樣。Yellowbrick的主要目標是創(chuàng)建一個和Scikit-Learn類似的有意義的API。其中最受歡迎的visualizers包括：
 
特征可視化
Rank Features: 對單個或者兩兩對應的特征進行排序以檢測其相關(guān)性
Parallel Coordinates: 對實例進行水平視圖
Radial Visualization: 在一個圓形視圖中將實例分隔開
PCA Projection: 通過主成分將實例投射
Feature Importances: 基于它們在模型中的表現(xiàn)對特征進行排序
Scatter and Joint Plots: 用選擇的特征對其進行可視化
分類可視化
Class Balance: 看類的分布怎樣影響模型
Classification Report: 用視圖的方式呈現(xiàn)精確率，召回率和F1值
ROC/AUC Curves: 特征曲線和ROC曲線子下的面積
Confusion Matrices: 對分類決定進行視圖描述
回歸可視化
Prediction Error Plot: 沿著目標區(qū)域?qū)δＰ瓦M行細分
Residuals Plot: 顯示訓練數(shù)據(jù)和測試數(shù)據(jù)中殘差的差異
Alpha Selection: 顯示不同alpha值選擇對正則化的影響
聚類可視化
K-Elbow Plot: 用肘部法則或者其他指標選擇k值
Silhouette Plot: 通過對輪廓系數(shù)值進行視圖來選擇k值
文本可視化
Term Frequency: 對詞項在語料庫中的分布頻率進行可視化
t-SNE Corpus Visualization: 用隨機鄰域嵌入來投射文檔

def testFunc1(savepath='Results/breast_cancer_ROCAUC.png'):
 '''
 基于癌癥數(shù)據(jù)集的測試
 '''
 data=load_breast_cancer()
 X,y=data['data'],data['target']
 X_train, X_test, y_train, y_test = train_test_split(X, y)
 viz=ROCAUC(LogisticRegression())
 viz.fit(X_train, y_train)
 viz.score(X_test, y_test)
 viz.poof(outpath=savepath)

def testFunc2(savepath='Results/breast_cancer_ParallelCoordinates.png'):
 '''
 用平行坐標的方法對高維數(shù)據(jù)進行作圖
 '''
 data=load_breast_cancer()
 X,y=data['data'],data['target']
 print 'X_shape: ',X.shape #X_shape: (569L, 30L)
 visualizer=ParallelCoordinates()
 visualizer.fit_transform(X,y)
 visualizer.poof(outpath=savepath)

def testFunc3(savepath='Results/breast_cancer_LR_report.png'):
 '''
 基于癌癥數(shù)據(jù)集，使用邏輯回歸模型來分類，繪制分類報告
 '''
 data=load_breast_cancer()
 X,y=data['data'],data['target']
 model=LogisticRegression()
 visualizer=ClassificationReport(model)
 X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.3,random_state=42)
 visualizer.fit(X_train,y_train)
 visualizer.score(X_test,y_test)
 visualizer.poof(outpath=savepath)

def testFunc5(savepath='Results/bikeshare_Rank2D.png'):
 '''
 共享單車數(shù)據(jù)集預測
 '''
 data=pd.read_csv('bikeshare/bikeshare.csv')
 X=data[["season", "month", "hour", "holiday", "weekday", "workingday",
   "weather", "temp", "feelslike", "humidity", "windspeed"
   ]]
 y=data["riders"]
 visualizer=Rank2D(algorithm="pearson")
 visualizer.fit_transform(X)
 visualizer.poof(outpath=savepath)

def testFunc7(savepath='Results/bikeshare_LinearRegression_ResidualsPlot.png'):
 '''
 基于共享單車數(shù)據(jù)使用線性回歸模型預測
 '''
 data = pd.read_csv('bikeshare/bikeshare.csv')
 X=data[["season", "month", "hour", "holiday", "weekday", "workingday",
   "weather", "temp", "feelslike", "humidity", "windspeed"]]
 y=data["riders"]
 X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.3)
 visualizer=ResidualsPlot(LinearRegression())
 visualizer.fit(X_train, y_train)
 visualizer.score(X_test, y_test)
 visualizer.poof(outpath=savepath)

def testFunc8(savepath='Results/bikeshare_RidgeCV_AlphaSelection.png'):
 '''
 基于共享單車數(shù)據(jù)使用AlphaSelection
 '''
 data=pd.read_csv('bikeshare/bikeshare.csv')
 X=data[["season", "month", "hour", "holiday", "weekday", "workingday",
   "weather", "temp", "feelslike", "humidity", "windspeed"]]
 y=data["riders"]
 alphas=np.logspace(-10, 1, 200)
 visualizer=AlphaSelection(RidgeCV(alphas=alphas))
 visualizer.fit(X, y)
 visualizer.poof(outpath=savepath)

def testFunc9(savepath='Results/bikeshare_Ridge_PredictionError.png'):
 '''
 基于共享單車數(shù)據(jù)繪制預測錯誤圖
 '''
 data=pd.read_csv('bikeshare/bikeshare.csv')
 X=data[["season", "month", "hour", "holiday", "weekday", "workingday",
   "weather", "temp", "feelslike", "humidity", "windspeed"]]
 y=data["riders"]
 X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.3)
 visualizer=PredictionError(Ridge(alpha=3.181))
 visualizer.fit(X_train, y_train)
 visualizer.score(X_test, y_test)
 visualizer.poof(outpath=savepath)
blog.csdn.net/Together_CZ/article/details/86640784