使用MinMaxScaler处理DataFrame时遇ValueError问题求助

在对小时级时序数据预处理以应用K-means聚类时，使用MinMaxScaler做归一化出现如下错误：

Traceback (most recent call last):
  File ".venv\lib\site-packages\pandas\core\series.py", line 191, in wrapper
    raise TypeError(f"cannot convert the series to {converter}")
TypeError: cannot convert the series to <class 'float'>

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File ".venv\timesequence.py", line 210, in <module>
    matrix = pd.DataFrame(scaler.fit_transform(x_calls), columns=df_hours.columns, index=df_hours.index)
  File ".venv\lib\site-packages\sklearn\base.py", line 867, in fit_transform
    return self.fit(X, **fit_params).transform(X)
  File ".venv\lib\site-packages\sklearn\preprocessing\_data.py", line 420, in fit
    return self.partial_fit(X, y)
  File ".venv\lib\site-packages\sklearn\preprocessing\_data.py", line 457, in partial_fit
    X = self._validate_data(
  File ".venv\lib\site-packages\sklearn\base.py", line 577, in _validate_data
    X = check_array(X, input_name="X", **check_params)
  File ".venv\lib\site-packages\sklearn\utils\validation.py", line 856, in check_array
    array = np.asarray(array, order=order, dtype=dtype)
  File ".venv\lib\site-packages\pandas\core\generic.py", line 2064, in __array__
    return np.asarray(self._values, dtype=dtype)
ValueError: setting an array element with a sequence.

预处理及归一化代码如下：

#--------------------Preprocessing ds 
counter_ = 0
zero = 0
df_hours = pd.DataFrame({
    'Hour': [],
    'SumView':[],
    'CountStudent':[]
}, dtype=object)

while counter_ < 24:
    if (counter_ in sub_data_hour['Hour']):
        row = sub_data_hour.loc[(pd.to_numeric(sub_data_hour['Hour'], errors='coerce')) == counter_]
        df_hours.loc[len(df_hours.index)] = [counter_, row['SumView'], row['CountStudent']]
    else:
        df_hours.loc[len(df_hours.index)] = [counter_, zero, zero]
    counter_ += 1

#----------Normalize dataset------------
x_calls = df_hours.columns[2:]
scaler = MinMaxScaler()
matrix = pd.DataFrame(scaler.fit_transform(df_hours[x_calls]), columns=x_calls, index=df_hours.index) 

错误原因 #

预处理循环中，当匹配到对应小时的行时，row['SumView']和row['CountStudent']返回的是Pandas Series对象（而非单个数值），导致df_hours的这两列混合了Series和标量值。MinMaxScaler要求输入为纯数值型二维数组，无法处理这种嵌套结构，因此抛出ValueError。

修复方案 #

1. 修正预处理逻辑，确保存入单个数值 #

将循环中赋值语句里的row['SumView']和row['CountStudent']改为取出Series的第一个元素（因为row是筛选后的单行DataFrame）：

2. 移除不必要的object类型指定 #

初始化DataFrame时不要强制dtype=object，让Pandas自动推断数值类型，避免后续类型转换问题。

修复后的完整代码 #

#--------------------Preprocessing ds 
counter_ = 0
zero = 0
# 不指定dtype=object，让Pandas自动推断数值类型
df_hours = pd.DataFrame({
    'Hour': [],
    'SumView': [],
    'CountStudent': []
})

while counter_ < 24:
    if counter_ in sub_data_hour['Hour']:
        # 筛选对应小时的行
        row = sub_data_hour.loc[pd.to_numeric(sub_data_hour['Hour'], errors='coerce') == counter_]
        # 用iloc[0]取出单个数值，而非Series对象
        df_hours.loc[len(df_hours.index)] = [counter_, row['SumView'].iloc[0], row['CountStudent'].iloc[0]]
    else:
        df_hours.loc[len(df_hours.index)] = [counter_, zero, zero]
    counter_ += 1

#----------Normalize dataset------------
x_calls = df_hours.columns[2:]
scaler = MinMaxScaler()
# 此时df_hours[x_calls]是纯数值型DataFrame，可直接传入fit_transform
matrix = pd.DataFrame(scaler.fit_transform(df_hours[x_calls]), columns=x_calls, index=df_hours.index) 

额外优化建议 #

可以用更高效的Pandas方法替代while循环，比如reindex来补全24小时的数据，避免手动循环：

# 将sub_data_hour的Hour列转为数值型
sub_data_hour['Hour'] = pd.to_numeric(sub_data_hour['Hour'], errors='coerce')
# 按Hour分组聚合
hourly_data = sub_data_hour.groupby('Hour')[['SumView', 'CountStudent']].sum()
# 重新索引补全0-23小时，缺失值填充0
df_hours = hourly_data.reindex(range(24), fill_value=0).reset_index().rename(columns={'index': 'Hour'})

这种方法代码更简洁，运行效率更高，也能避免手动循环带来的错误。

内容的提问来源于stack exchange，提问作者teresaroserain