Index de l'article

Work with dates

Group by months

If you have a data-frame df with a field Created at containing dates like:

2021-02-05 10:34:21 UTC

You can work with it, example to group and sort by month, you can cheat doing:

df_Temp = pd.DataFrame(df['Created at'].str[:7])
df_Created_count = pd.DataFrame(df_Temp.groupby(['Created at'], dropna=False).size(), columns=['Total']).sort_values(['Created at'], ascending=True).reset_index()

indeed with str[:7] we extract years and months from strings. Then a usual groupby.

Ok but it is more a pandas-way to convert your field in a true date format, in a new virtual field. Then you can extract real months and store them in a proper data-frame.

df['Created'] = pd.to_datetime(df['Created at'])
df['Created'] = df['Created'].dt.to_period("M")
df_TempCreated = pd.DataFrame(df['Created'])

Then easy to group and count them.

df_Created_count = pd.DataFrame(df_TempCreated.groupby(['Created'], dropna=False).size(), columns=['Total']).sort_values(['Created'], ascending=True).reset_index()

As now you use a real date format, you can display the full months with dt.strftime.

df_Created_count['Created'] = df_Created_count['Created'].dt.strftime('%B %Y')

Extract the previous month

PreviousMonth = today - pd.DateOffset(months=1)