CODEWITHSUNDEEP
pythonpandasdataframeparsing
tonycor nikolauos
tonycor nikolauos

Reputation: 475

Convert pandas column of Object dtype into time datetime (with micro seconds)

My df looks like this. I tried many ways to convert the time of: hour:minute:second microseconds, but hit a wall. Below is the sample.df.

I need the Time and micro-seconds (the 3 numbers after 00:00:02 948, so the micro seconds in this row would be: 948.

How to make another column of Time object that pandas will be able to work- like with fast vectorizing (not iterating) later on. I do not need year, month, day, date. Just need theTime, with microseconds, formatted in a way to be acceptable for work by pd, like making aggregated time intervals etc. Any ideas would be greatly appreciated!

idx,   time,   col2, col3,  col4, col5,  col6 
0   00:00:02 948    41   14   34      36    55
1   00:00:03 228    44  -12   32      41    55

Upvotes: 1

Views: 1069

Answers (1)

bitflip
bitflip

Reputation: 3684

You only have a "timedelta", not a datetime. So by converting it to datetime, pandas assumes time after 1900-01-01, because thats the definition (timestamp of 0).

I did not completely understand how your output is supposed to look like, so feel free to ask how to change it if I misunderstood.

You can access attributes for minute/second/microsecond independently and format them like this:

import pandas as pd
df = pd.DataFrame(columns=["time"], data=["00:00:02 948", "00:00:03 228"])
df["time"] = pd.to_datetime(df["time"], format='%H:%M:%S %f').apply(lambda dt: f"{dt.minute:02d}:{dt.second:02d}.{dt.microsecond:06d}")

outputs:

           time
0  00:02.948000
1  00:03.228000

If you only want to get seconds and microseconds, just remove dt.minute from the formatting.

pd.to_datetime(df["time"], format='%H:%M:%S %f').apply(lambda dt: f"{dt.second:02d}.{dt.microsecond:06d}")

        time
0  02.948000
1  03.228000

Upvotes: 1

Related Questions