python - 如何正确使用 sklearn 的 cross_validate 和 One Hot Encoded 类？

Question

我创建了一个模型来对我的 8 类数据集进行分类，并使用 MLP 从中获得一些分数。为此，我决定使用 sklearn.metrics.cross_validate，使用 10 折。

以下代码工作正常：

from sklearn.neural_network import MLPClassifier
from sklearn.model_selection import cross_validate
from sklearn.metrics import accuracy_score, make_scorer, f1_score
import pandas as pd

def MLPClasify(sample):
    df = pd.read_csv('my_path\\my_file.csv', header=None)
    y = df[NumberOfFeatures]
    x = df.drop([NumberOfFeatures], axis=1)
    clf = MLPClassifier(hidden_layer_sizes=(27), activation='logistic', max_iter=500, alpha=0.0001, 
                        solver='adam', verbose=10, random_state=21, tol=0.000000001)
    clf.out_activation_ = 'softmax'
    scoring = {'Accuracy': make_scorer(accuracy_score), 'F1': make_scorer(f1_score, 
               average='weighted')}
    scores = cross_validate(clf, x, y, cv=10, scoring=scoring)
    return scores

一切顺利，我得到了大约 60% 的准确度。所以我决定使用一种热编码，看看能不能得到更好的结果。所以我写了以下代码：

from sklearn.neural_network import MLPClassifier
from sklearn.model_selection import cross_validate
from sklearn.preprocessing import LabelEncoder, OneHotEncoder
from sklearn.metrics import accuracy_score, make_scorer, f1_score
import pandas as pd

def MLPClasify(sample):
    df = pd.read_csv('my_path\\my_file.csv', header=None)
    y = df[NumberOfFeatures]
    x = df.drop([NumberOfFeatures], axis=1)
    label_encoder = LabelEncoder()
    integer_encoded = label_encoder.fit_transform(y)
    onehot_encoder = OneHotEncoder()
    integer_encoded = integer_encoded.reshape(len(integer_encoded), 1)
    onehot_encoded = onehot_encoder.fit_transform(integer_encoded)
    y = onehot_encoded
    clf = MLPClassifier(hidden_layer_sizes=(27), activation='logistic', max_iter=500, alpha=0.0001, 
                        solver='adam', verbose=10, random_state=21, tol=0.000000001)
    clf.out_activation_ = 'softmax'
    scoring = {'Accuracy': make_scorer(accuracy_score), 'F1': make_scorer(f1_score, 
               average='weighted')}
    scores = cross_validate(clf, x, y, cv=10, scoring=scoring)
    return scores

好吧，代码运行，但我收到以下警告：

UndefinedMetricWarning：F-score 定义不明确，在没有真实样本和预测样本的标签中设置为 0.0。使用zero_division参数来控制这种行为。平均，“真实或预测”，“F 分数是”，len(true_sum)

此外，我的准确率下降到不到 2%

关于我可能做错了什么的任何想法？

谢谢您的帮助

标签： pythonscikit-learncross-validationone-hot-encoding