如何在交叉验证和 GridSearchCV 中实现 SMOTE

Question

我对 Python 比较陌生。您能帮我改进 SMOTE 的实施，使其成为合适的管道吗？我想要的是在每次 k 次迭代的训练集上应用过采样和欠采样，以便模型在平衡数据集上进行训练，并在不平衡的遗漏部分上进行评估。问题是，当我这样做时，我无法使用熟悉的 sklearn 界面进行评估和网格搜索。

是否可以制作类似于model_selection.RandomizedSearchCV的东西。我对此的看法：

df = pd.read_csv("Imbalanced_data.csv") #Load the data set
X = df.iloc[:,0:64]
X = X.values
y = df.iloc[:,64]
y = y.values
n_splits = 2
n_measures = 2 #Recall and AUC
kf = StratifiedKFold(n_splits=n_splits) #Stratified because we need balanced samples
kf.get_n_splits(X)
clf_rf = RandomForestClassifier(n_estimators=25, random_state=1)
s =(n_splits,n_measures)
scores = np.zeros(s)
for train_index, test_index in kf.split(X,y):
   print("TRAIN:", train_index, "TEST:", test_index)
   X_train, X_test = X[train_index], X[test_index]
   y_train, y_test = y[train_index], y[test_index]
   sm = SMOTE(ratio = 'auto',k_neighbors = 5, n_jobs = -1)
   smote_enn = SMOTEENN(smote = sm)
   x_train_res, y_train_res = smote_enn.fit_sample(X_train, y_train)
   clf_rf.fit(x_train_res, y_train_res)
   y_pred = clf_rf.predict(X_test,y_test)
   scores[test_index,1] = recall_score(y_test, y_pred)
   scores[test_index,2] = auc(y_test, y_pred)

Answer 1

这看起来符合要求http://contrib.scikit-learn.org/imbalanced-learn/stable/generated/imblearn.over_sampling.SMOTE.html

您需要创建自己的 transformer (http://scikit-learn.org/stable/modules/generated/sklearn.base.TransformerMixin.html)，调用 fit returns 平衡数据集（大概是从 StratifiedKFold)，但在调用 predict 时，测试数据将发生这种情况，调用 SMOTE。

Answer 2

您需要查看管道对象。不平衡学习有一个 Pipeline 扩展了 scikit-learn 管道，除了 fit_predict()，fit_transform 之外还适应 fit_sample() 和 sample() 方法scikit-learn 的 () 和 predict() 方法。

在此处查看此示例：

https://imbalanced-learn.org/stable/auto_examples/pipeline/plot_pipeline_classification.html

对于您的代码，您可能希望这样做：

from imblearn.pipeline import make_pipeline, Pipeline

smote_enn = SMOTEENN(smote = sm)
clf_rf = RandomForestClassifier(n_estimators=25, random_state=1)

pipeline = make_pipeline(smote_enn, clf_rf)
    OR
pipeline = Pipeline([('smote_enn', smote_enn),
                     ('clf_rf', clf_rf)])

然后您可以将此pipeline对象作为常规对象传递给scikit-learn中的GridSearchCV、RandomizedSearchCV或其他交叉验证工具。

kf = StratifiedKFold(n_splits=n_splits)
random_search = RandomizedSearchCV(pipeline, param_distributions=param_dist,
                                   n_iter=1000, 
                                   cv = kf)

如何在交叉验证和 GridSearchCV 中实现 SMOTE

How to implement SMOTE in cross validation and GridSearchCV

python

scikit-learn

pipeline

cross-validation

grid-search