使用 Python 构建自定义 RandomSearchCV

Question

我正在尝试从头开始构建自定义 K 折 RandomSearchCV。我了解 RandomSearchCV 的工作原理，我正在尝试在随机生成的数据集上从头开始实现它。当我尝试运行代码时，出现以下错误。我认为这与我在x_train 列表中创建组的方式有关。这个错误是什么及其修复？ ：

ValueError                                Traceback (most recent call last)
<ipython-input-12-229cc493eeb9> in <module>
     41 
     42 classifier = KNeighborsClassifier()
---> 43 RandomSearchCV(X_train,y_train, classifier, folds = 3)
     44 
     45 

<ipython-input-12-229cc493eeb9> in RandomSearchCV(x_train, y_train, classifier, folds)
     26             #classifier (K-NN)
     27             classifier.n_neighbors = parameter
---> 28             classifier.fit(x_train_group, y_train_group)
     29 
     30             #Predicton

~\anaconda3\lib\site-packages\sklearn\neighbors\_base.py in fit(self, X, y)
   1128         """
   1129         if not isinstance(X, (KDTree, BallTree)):
-> 1130             X, y = check_X_y(X, y, "csr", multi_output=True)
   1131 
   1132         if y.ndim == 1 or y.ndim == 2 and y.shape[1] == 1:

~\anaconda3\lib\site-packages\sklearn\utils\validation.py in check_X_y(X, y, accept_sparse, accept_large_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, multi_output, ensure_min_samples, ensure_min_features, y_numeric, warn_on_dtype, estimator)
    753                     ensure_min_features=ensure_min_features,
    754                     warn_on_dtype=warn_on_dtype,
--> 755                     estimator=estimator)
    756     if multi_output:
    757         y = check_array(y, 'csr', force_all_finite=True, ensure_2d=False,

~\anaconda3\lib\site-packages\sklearn\utils\validation.py in check_array(array, accept_sparse, accept_large_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, warn_on_dtype, estimator)
    572         if not allow_nd and array.ndim >= 3:
    573             raise ValueError("Found array with dim %d. %s expected <= 2."
--> 574                              % (array.ndim, estimator_name))
    575 
    576         if force_all_finite:

ValueError: Found array with dim 3. Estimator expected <= 2.

这是我的实现：

from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
import numpy
from tqdm import tqdm
import numpy as np
from sklearn.metrics.pairwise import euclidean_distances
import random
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import accuracy_score


x,y = make_classification(n_samples=10000, n_features=2, n_informative=2, n_redundant= 0, n_clusters_per_class=1, random_state=60)
X_train, X_test, y_train, y_test = train_test_split(x,y,stratify=y,random_state=42)

def RandomSearchCV(x_train,y_train, classifier, folds):
    train_scores = []
    test_scores = []

    #1. Generating 10 unique values from given range
    params = random.sample(range(0, 50), 10)
    x_train_split = []
    y_train_split = []
    #dividing x_train into groups
    for i in range(0, len(x_train), int(len(x_train)/folds)):
        x_train_split.append(x_train[i:i+int(len(x_train)/folds)])
        y_train_split.append(y_train[i:i+int(len(y_train)/folds)])

    #3.for each hyperparameter that we generated in step 1 and dividing dataset into training and CV datasets:

    for parameter in params:
        trainscores_folds = []
        testscores_folds  = []

        for group in range(len(x_train_split)):
            x_train_group = x_train_split[0:group] + x_train_split[group+1:]
            x_cv_group = [x_train_split[group]]
            y_train_group = y_train_split[0:group] + y_train_split[group+1:]
            y_cv_group = [y_train_split[group]]

            #classifier (K-NN)
            classifier.n_neighbors = parameter
            classifier.fit(x_train_group, y_train_group)

            #Predicton
            y_pred = classifier.predict(x_cv_group)
            testscores_folds.append(accuracy_score(y_cv_group, Y_pred))

            y_pred = classifier.predict(x_train_group)
            trainscores_folds.append(accuracy_score(y_train_group, Y_pred))

        trainscores.append(np.mean(np.array(trainscores_folds)))
        testscores.append(np.mean(np.array(testscores_folds)))

    return trainscores, testscores

classifier = KNeighborsClassifier()
RandomSearchCV(X_train,y_train, classifier, folds = 3)

感谢您的帮助。

Answer 1

由于错误状态，您使用的是一个三维数组，而 classifier.fit() 方法需要一个二维矩阵。解决此问题所需要做的就是将您的 train/cv/test 组更改为如下所示：

for group in range(len(x_train_split)):
    x_train_group = np.concatenate(x_train_split[0:group] + x_train_split[group+1:])
    x_cv_group = x_train_split[group]
    y_train_group = np.concatenate(y_train_split[0:group] + y_train_split[group+1:])
    y_cv_group = y_train_split[group]
    ....

Answer 2

x_train_group 是一个数组列表，这使它成为 3 维的（如错误中所述）。这不适用于拟合分类器，因为它需要二维输入。尝试调用np.concatenate(x_train_group) 连接折叠并使其成为二维输入。