how do you convert pandas data frame epoch values to readable datetime format

I have a data frame like this:

print(df)
Metric  TimeStamp       Value
cpu     1642039200000   4
cpu     1642042800000   3
cpu     1642046400000   5

I need to convert TimeStamp to readaable time stamp yearmonthdate hour:min:sec

I have tried:

import datetime

df['TimeStamp']=datetime.datetime.fromtimestamp(df['TimeStamp'])

does not seem to be working.

>Solution :

datetime.datetime.fromtimestamp expects a single value. To work with series, use pandas’ methods:

pd.to_datetime(df['TimeStamp']*1000000, utc=True).dt.tz_convert('EST').dt.strftime("%Y-%m-%d")

Example:

df = pd.DataFrame({'ts': [1642039200000, 1642042800000, 1642046400000]}) 
df['ts'] = pd.to_datetime(df['ts']*1000000, utc=True).dt.tz_convert('EST')
df['ts']

Output:

0   2022-01-12 21:00:00-05:00
1   2022-01-12 22:00:00-05:00
2   2022-01-12 23:00:00-05:00
Name: ts, dtype: datetime64[ns, EST]

To get a specific string format, df['ts'].dt.strftime("%Y-%m-%d") will output:

0    2022-01-12
1    2022-01-12
2    2022-01-12
Name: ts, dtype: object