Pandas重采样触发ValueError: a must be greater than 0问题求助

问题背景 #

尝试对重平衡数据集churn_train中Churn列值为True的样本重采样，抽取158条记录，但触发报错。已确认数据集非空，Churn为True的样本共320条。

相关信息 #

数据集片段（churn部分行） #

State,Account Length,Area Code,Phone,Intl Plan,VMail Plan,VMail Message,Day Mins,Day Calls,Day Charge,Eve Mins,Eve Calls,Eve Charge,Night Mins,Night Calls,Night Charge,Intl Mins,Intl Calls,Intl Charge,CustServ Calls,Old Churn,Churn
"KS",128,415,"382-4657","no","yes",25,265.100000,110,45.070000,197.400000,99,16.780000,244.700000,91,11.010000,10.000000,3,2.700000,1,"False.","False"
"OH",107,415,"371-7191","no","yes",26,161.600000,123,27.470000,195.500000,103,16.620000,254.400000,103,11.450000,13.700000,3,3.700000,1,"False.","False"
"IN",65,415,"329-6603","no","no",0,129.100000,137,21.950000,228.500000,83,19.420000,208.800000,111,9.400000,12.700000,6,3.430000,4,"True.","True"

执行代码 #

churn_train['Churn'].value_counts()
# 输出：
# False    1913
# True      320
# Name: Churn, dtype: int64

to_resample = churn_train.loc[churn_train['Churn'] == "True"]
our_resample = to_resample.sample(n = 158, replace = True)
churn_train_rebal = pd.concat([churn_train, our_resample])

报错信息 #

ValueError                                Traceback (most recent call last)
/var/folders/wv/42dn23fd1cb0czpvqdnb6zw00000gn/T/ipykernel_7751/2929105044.py in <module>
      1 to_resample = churn_train.loc[churn_train['Churn'] == "True"]
----> 2 our_resample = to_resample.sample(n = 158, replace = True)
      3 churn_train_rebal = pd.concat([churn_train, our_resample])

~/opt/miniconda3/lib/python3.9/site-packages/pandas/core/generic.py in sample(self, n, frac, replace, weights, random_state, axis, ignore_index)
   5452             weights = sample.preprocess_weights(self, weights, axis)
   5453 
-> 5454         sampled_indices = sample.sample(obj_len, size, replace, weights, rs)
   5455         result = self.take(sampled_indices, axis=axis)
   5456 

~/opt/miniconda3/lib/python3.9/site-packages/pandas/core/sample.py in sample(obj_len, size, replace, weights, random_state)
    148             raise ValueError("Invalid weights: weights sum to zero")
    149 
-> 150     return random_state.choice(obj_len, size=size, replace=replace, p=weights).astype(
    151         np.intp, copy=False
    152     )

mtrand.pyx in numpy.random.mtrand.RandomState.choice()

ValueError: a must be greater than 0 unless no samples are taken

问题原因 #

从value_counts的输出可以看到，Churn列的取值是布尔类型的True/False，但代码中用了字符串"True"进行匹配，导致筛选出的to_resample是空DataFrame。空DataFrame调用sample(n=158)时，就会触发ValueError——因为没有可采样的样本。

解决方法 #

将筛选条件修改为匹配布尔值True，而非字符串"True"：

# 方法1：直接匹配布尔值
to_resample = churn_train.loc[churn_train['Churn'] == True]
# 方法2：更简洁的布尔索引写法
to_resample = churn_train[churn_train['Churn']]

our_resample = to_resample.sample(n = 158, replace = True)
churn_train_rebal = pd.concat([churn_train, our_resample])

如果不确定列的数据类型，可以先执行print(churn_train['Churn'].dtype)确认类型，再调整匹配条件。

内容的提问来源于stack exchange，提问作者c200402