python中文本分类的过采样？答案

【问题标题】：Oversampling for text classification in python?python中文本分类的过采样？
【发布时间】：2020-09-17 19:01:58
【问题描述】：

我有一个要分类的文本数据框。但我需要先进行过采样。请在下面找到示例数据：

df=[['I am going to class today','I am going to class today','I am going to class today','I am going to class today','I am going to class today','I am going to class today','I am going to class today','I am going to class today','I am going to class today','I am going to class today','I am not going to class today','I am not going to class today','I am not going to class today','I am not going to class today'],['Positive','Positive','Positive','Positive','Positive','Positive','Positive','Positive','Positive','Positive','Negative','Negative','Negative','Negative']]
df=pd.DataFrame(df)
df=df.transpose()
df.columns=['Features','Class']
df
          Features                       Class
0   I am going to class today       Positive
1   I am going to class today       Positive
2   I am going to class today       Positive
3   I am going to class today       Positive
4   I am going to class today       Positive
5   I am going to class today       Positive
6   I am going to class today       Positive
7   I am going to class today       Positive
8   I am going to class today       Positive
9   I am going to class today       Positive
10  I am not going to class today   Negative
11  I am not going to class today   Negative
12  I am not going to class today   Negative
13  I am not going to class today   Negative

oversample = RandomOverSampler(sampling_strategy='minority')
# fit and apply the transform
X_over, y_over = oversample.fit_resample(df['Features'], df['Class'])
# summarize class distribution
print(Counter(y_over))

但这不起作用并给我ValueError: Expected 2D array, got 1D array instead:。如何对这些数据进行过采样？

【问题讨论】：

请指定进口

标签： python machine-learning scikit-learn oversampling

【解决方案1】：

我发现了问题。我需要重塑我的数据。

X_over, y_over = oversample.fit_resample(df['Features'].values.reshape(-1,1), df['Class'])

这正在工作。

Counter({'Positive': 10, 'Negative': 10})

【讨论】：

您可以简单地使用df[['Features']] 代替df['Features]。