ValueError:zerosize数组到没有标识的缩减操作fmin

Question

我有下面的代码

def order(frame, feature):
    # Enter the dataset and the list of features that will need to be ordered
    # !!!! the list of feture must be qualitative data
    # Return:
    # 1. the dataset with the feature ordered against the target label
    # 2. the list of qualitative feature name ordered
    frame_cop = frame.copy()
    qual_encoded = []
    for f in feature:
        ordering = pd.DataFrame()
        ordering['val'] = frame_cop[f].unique()
        ordering.index = ordering.val
        ordering['spmean'] = frame_cop[[f, 'SalePrice']].groupby(f).mean()['SalePrice']
        ordering = ordering.sort_values('spmean')
        ordering['ordering'] = range(1, ordering.shape[0]+1)
        ordering = ordering['ordering'].to_dict()
        for cat, o in ordering.items():
            frame_cop.loc[frame_cop[f] == cat, f] = o
        qual_encoded.append(f)
    return frame_cop, qual_encoded

当我想可视化热图时，如下所示：

plt.figure()
train_csv_up, qual_order = order(train_csv, qualitative_card)
sns.heatmap(train_csv_up[qual_order].corr())

我有错误消息：ValueError:zero size数组到没有标识的缩减操作fmin

或者，如果我添加以下['SalePrice']

plt.figure()
train_csv_up, qual_order = order(train_csv, qualitative_card)
sns.heatmap(train_csv_up[qual_order+['SalePrice'].corr())

热图仅显示售价

但是，如果我更改下面的这些行

for cat, o in ordering.items():
    frame_cop.loc[frame_cop[f] == cat, f] = o
qual_encoded.append(f)

到

for cat, o in ordering.items():
    frame_cop.loc[frame_cop[f] == cat, f+'_q'] = o
qual_encoded.append(f+'_q')

然后它就可以工作了（有和没有SalePrice），但我想避免有一个新的特性名称列表，并保留f而不是f+“”q'

但我不明白为什么它在一种情况下有效，而在另一种情况下无效

期待阅读您的评论

问候