利用scikitlearn畫ROC曲線例項

阿新 • • 發佈：2020-07-03

一個完整的資料探勘模型，最後都要進行模型評估，對於二分類來說，AUC，ROC這兩個指標用到最多，所以利用sklearn裡面相應的函式進行模組搭建。

具體實現的程式碼可以參照下面博友的程式碼，評估svm的分類指標。注意裡面的一些細節需要注意，一個是呼叫roc_curve 方法時，指明目標標籤，否則會報錯。

具體是這個引數的設定pos_label ，以前在unionbigdata實習時學到的。

重點是以下的程式碼需要根據實際改寫：

  mean_tpr = 0.0 
  mean_fpr = np.linspace(0,1,100) 
  all_tpr = []
  
  y_target = np.r_[train_y,test_y]
  cv = StratifiedKFold(y_target,n_folds=6)
 
    #畫ROC曲線和計算AUC
    fpr,tpr,thresholds = roc_curve(test_y,predict,pos_label = 2)##指定正例標籤，pos_label = ###########在數之聯的時候學到的，要制定正例
    
    mean_tpr += interp(mean_fpr,fpr,tpr)     #對mean_tpr在mean_fpr處進行插值，通過scipy包呼叫interp()函式 
    mean_tpr[0] = 0.0                #初始處為0 
    roc_auc = auc(fpr,tpr) 
    #畫圖，只需要plt.plot(fpr,tpr),變數roc_auc只是記錄auc的值，通過auc()函式能計算出來 
    plt.plot(fpr,lw=1,label='ROC %s (area = %0.3f)' % (classifier,roc_auc))

然後是博友的參考程式碼：

# -*- coding: utf-8 -*- 
""" 
Created on Sun Apr 19 08:57:13 2015 
@author: shifeng 
""" 
print(__doc__) 
 
import numpy as np 
from scipy import interp 
import matplotlib.pyplot as plt 
 
from sklearn import svm,datasets 
from sklearn.metrics import roc_curve,auc 
from sklearn.cross_validation import StratifiedKFold 
 
############################################################################### 
# Data IO and generation,匯入iris資料，做資料準備 
 
# import some data to play with 
iris = datasets.load_iris() 
X = iris.data 
y = iris.target 
X,y = X[y != 2],y[y != 2]#去掉了label為2，label只能二分，才可以。 
n_samples,n_features = X.shape 
 
# Add noisy features 
random_state = np.random.RandomState(0) 
X = np.c_[X,random_state.randn(n_samples,200 * n_features)] 
 
############################################################################### 
# Classification and ROC analysis 
#分類，做ROC分析 
 
# Run classifier with cross-validation and plot ROC curves 
#使用6折交叉驗證，並且畫ROC曲線 
cv = StratifiedKFold(y,n_folds=6) 
classifier = svm.SVC(kernel='linear',probability=True,random_state=random_state)#注意這裡，probability=True,需要，不然預測的時候會出現異常。另外rbf核效果更好些。 
mean_tpr = 0.0 
mean_fpr = np.linspace(0,100) 
all_tpr = [] 
 
for i,(train,test) in enumerate(cv): 
  #通過訓練資料，使用svm線性核建立模型，並對測試集進行測試，求出預測得分 
  probas_ = classifier.fit(X[train],y[train]).predict_proba(X[test]) 
#  print set(y[train])           #set([0,1]) 即label有兩個類別 
#  print len(X[train]),len(X[test])    #訓練集有84個，測試集有16個 
#  print "++",probas_           #predict_proba()函式輸出的是測試集在lael各類別上的置信度， 
#  #在哪個類別上的置信度高，則分為哪類 
  # Compute ROC curve and area the curve 
  #通過roc_curve()函式，求出fpr和tpr，以及閾值 
  fpr,thresholds = roc_curve(y[test],probas_[:,1]) 
  mean_tpr += interp(mean_fpr,tpr)     #對mean_tpr在mean_fpr處進行插值，通過scipy包呼叫interp()函式 
  mean_tpr[0] = 0.0                #初始處為0 
  roc_auc = auc(fpr,tpr) 
  #畫圖，只需要plt.plot(fpr,變數roc_auc只是記錄auc的值，通過auc()函式能計算出來 
  plt.plot(fpr,label='ROC fold %d (area = %0.2f)' % (i,roc_auc)) 
 
#畫對角線 
plt.plot([0,1],[0,'--',color=(0.6,0.6,0.6),label='Luck') 
 
mean_tpr /= len(cv)           #在mean_fpr100個點，每個點處插值插值多次取平均 
mean_tpr[-1] = 1.0           #座標最後一個點為（1,1） 
mean_auc = auc(mean_fpr,mean_tpr)   #計算平均AUC值 
#畫平均ROC曲線 
#print mean_fpr,len(mean_fpr) 
#print mean_tpr 
plt.plot(mean_fpr,mean_tpr,'k--',label='Mean ROC (area = %0.2f)' % mean_auc,lw=2) 
 
plt.xlim([-0.05,1.05]) 
plt.ylim([-0.05,1.05]) 
plt.xlabel('False Positive Rate') 
plt.ylabel('True Positive Rate') 
plt.title('Receiver operating characteristic example') 
plt.legend(loc="lower right") 
plt.show()

補充知識：批量進行One-hot-encoder且進行特徵欄位拼接，並完成模型訓練demo

import org.apache.spark.ml.Pipeline
import org.apache.spark.ml.feature.{StringIndexer,OneHotEncoder}
import org.apache.spark.ml.feature.VectorAssembler
import ml.dmlc.xgboost4j.scala.spark.{XGBoostEstimator,XGBoostClassificationModel}
import org.apache.spark.ml.evaluation.BinaryClassificationEvaluator
import org.apache.spark.ml.tuning.{ParamGridBuilder,CrossValidator}
import org.apache.spark.ml.PipelineModel
 
val data = (spark.read.format("csv")
 .option("sep",",")
 .option("inferSchema","true")
 .option("header","true")
 .load("/Affairs.csv"))
 
data.createOrReplaceTempView("res1")
val affairs = "case when affairs>0 then 1 else 0 end as affairs,"
val df = (spark.sql("select " + affairs +
 "gender,age,yearsmarried,children,religiousness,education,occupation,rating" +
 " from res1 "))
 
val categoricals = df.dtypes.filter(_._2 == "StringType") map (_._1)
val indexers = categoricals.map(
 c => new StringIndexer().setInputCol(c).setOutputCol(s"${c}_idx")
)
 
val encoders = categoricals.map(
 c => new OneHotEncoder().setInputCol(s"${c}_idx").setOutputCol(s"${c}_enc").setDropLast(false)
)
 
val colArray_enc = categoricals.map(x => x + "_enc")
val colArray_numeric = df.dtypes.filter(_._2 != "StringType") map (_._1)
val final_colArray = (colArray_numeric ++ colArray_enc).filter(!_.contains("affairs"))
val vectorAssembler = new VectorAssembler().setInputCols(final_colArray).setOutputCol("features")
 
/*
val pipeline = new Pipeline().setStages(indexers ++ encoders ++ Array(vectorAssembler))
pipeline.fit(df).transform(df)
*/
 
///
// Create an XGBoost Classifier 
val xgb = new XGBoostEstimator(Map("num_class" -> 2,"num_rounds" -> 5,"objective" -> "binary:logistic","booster" -> "gbtree")).setLabelCol("affairs").setFeaturesCol("features")
 
// XGBoost paramater grid
val xgbParamGrid = (new ParamGridBuilder()
  .addGrid(xgb.round,Array(10))
  .addGrid(xgb.maxDepth,Array(10,20))
  .addGrid(xgb.minChildWeight,Array(0.1))
  .addGrid(xgb.gamma,Array(0.1))
  .addGrid(xgb.subSample,Array(0.8))
  .addGrid(xgb.colSampleByTree,Array(0.90))
  .addGrid(xgb.alpha,Array(0.0))
  .addGrid(xgb.lambda,Array(0.6))
  .addGrid(xgb.scalePosWeight,Array(0.1))
  .addGrid(xgb.eta,Array(0.4))
  .addGrid(xgb.boosterType,Array("gbtree"))
  .addGrid(xgb.objective,Array("binary:logistic")) 
  .build())
 
// Create the XGBoost pipeline
val pipeline = new Pipeline().setStages(indexers ++ encoders ++ Array(vectorAssembler,xgb))
 
// Setup the binary classifier evaluator
val evaluator = (new BinaryClassificationEvaluator()
  .setLabelCol("affairs")
  .setRawPredictionCol("prediction")
  .setMetricName("areaUnderROC"))
 
// Create the Cross Validation pipeline,using XGBoost as the estimator,the
// Binary Classification evaluator,and xgbParamGrid for hyperparameters
val cv = (new CrossValidator()
  .setEstimator(pipeline)
  .setEvaluator(evaluator)
  .setEstimatorParamMaps(xgbParamGrid)
  .setNumFolds(3)
  .setSeed(0))
 
 // Create the model by fitting the training data
val xgbModel = cv.fit(df)
 
 // Test the data by scoring the model
val results = xgbModel.transform(df)
 
// Print out a copy of the parameters used by XGBoost,attention pipeline
(xgbModel.bestModel.asInstanceOf[PipelineModel]
 .stages(5).asInstanceOf[XGBoostClassificationModel]
 .extractParamMap().toSeq.foreach(println))
results.select("affairs","prediction").show
 
println("---Confusion Matrix------")
results.stat.crosstab("affairs","prediction").show()
 
// What was the overall accuracy of the model,using AUC
val auc = evaluator.evaluate(results)
println("----AUC--------")
println("auc="+auc)

以上這篇利用scikitlearn畫ROC曲線例項就是小編分享給大家的全部內容了，希望能給大家一個參考，也希望大家多多支援我們。