CODEWITHSUNDEEP
pythonpandas
Theo75
Theo75

Reputation: 487

datetime columns losts time information after convertion to datetime

Please consider a panda dataframe df created with pd.read_csv() function.

df has 2 columns with type below:

df.dtypes:
DAT_RUN      object
DAT_FORECAST object

Some values:

DAT_RUN     DAT_FORECAST
0   2020-08-11 00:00:00.000     2020-08-11 00:00:00.000
1   2020-08-11 00:00:00.000     2020-08-11 01:00:00.000
2   2020-08-11 00:00:00.000     2020-08-11 02:00:00.000
3   2020-08-11 00:00:00.000     2020-08-11 03:00:00.000
4   2020-08-11 00:00:00.000     2020-08-11 04:00:00.000
...     ...     ...

As you can see, columns values are in date format.

  1. I want to convert theses columns to datetime (NOT IN PLACE):

    pd.to_datetime(df['DAT_RUN'], format='%Y-%m-%d %H:%M:%S')

enter image description here

I lost time information in value.

whereas

pd.to_datetime(df['DAT_FORECAST'], format='%Y-%m-%d %H:%M:%S')

enter image description here

keeps time information in value.

Why?

  1. And also, I want to search rows with restrictions:

For example in csv file I have:

enter image description here

df.loc[df['DAT_RUN'] == "2020-08-10 03:00:00.000", "DAT_RUN"]

returns 0 rows:

Series([], Name: DAT_RUN, dtype: object)

Whereas

df.loc[df['DAT_FORECAST'] == "2021-06-11 06:00:00.000", "DAT_FORECAST"]

returns rows:

enter image description here

What are the differences?

Upvotes: 0

Views: 38

Answers (1)

jezrael
jezrael

Reputation: 862791

If there is time with 00:00:00.000 pandas only not display it.

df['DAT_RUN'] = pd.to_datetime(df['DAT_RUN'])
df['DAT_FORECAST'] = pd.to_datetime(df['DAT_FORECAST'])

Check it:

print (df['DAT_RUN'].head().tolist())

For second issue seems are compared object columns instead datetimes, maybe not assigned back columns.

Upvotes: 1

Related Questions