pandas.to_datetime：类型错误：类型提升无效

Question

I get

TypeError: invalid type promotion

at

df.iloc[1:,0] = pd.to_datetime(np.where(m, df.iloc[1:,0], np.NaN), errors='ignore')

in the below code.

It is because at df.iloc[1, 0], the empty value has m = True. How do I solve this issue?

m = df.iloc[1:,0].apply(lambda v: isinstance(v, datetime))
df1 ['Date'] = df.iloc[1:,0]
df1 ['m'] = m
print(df1)

df.iloc[1:,0] = pd.to_datetime(np.where(m, df.iloc[1:,0], np.NaN), errors='ignore')

**OUTPUT of df1 **

|Date            |  m   |
|                | TRUE |
|2000-01-31      | TRUE |

Alternative
I tested the below as well, but I still get same error at

df.iloc[1:,0] = pd.to_datetime(np.where(m, df.iloc[1:,0], np.NaN), errors='ignore')

Regardless if errors='ignore' or errors='coerce'

m = df.iloc[1:,0].apply(lambda v: isinstance(v, datetime) if pd.isnull(v)==False else False)

df1 ['Date'] = df.iloc[1:,0]
df1 ['m'] = m
print(df1)

df.iloc[1:,0] = pd.to_datetime(np.where(m, df.iloc[1:,0], np.NaN), errors='ignore')

**OUTPUT of df1 **

|Date            |  m   |
|                | FALSE |
|2000-01-31      | TRUE |

Answer 1

The column was read in as datetime. Hence np.NaN led to the error in

np.where(m, df.iloc[1:,0], np.NaN)

Changing to np.datetime64('NaT') , solved the issue