sugestions on how to speed up lambda function-CodePudding

Any suggestions on how to speed up this little code?

The code works fine, it's just too slow.

 df['business_day'] = df['date'].apply(lambda x: isbday(x, holidays=holidays.US()))

My goal is to have a column that flags (true or false) if a given date is a business day, considering US holidays

CodePudding user response：

You're constructing the US instance again and again for every row you're checking. Try constructing it just once, eg:

us_holidays = holidays.US()
df['business_day'] = df['date'].apply(lambda x: isbday(x, holidays=us_holidays))

I'm not sure how much of an improvement you'll get time-wise but it'll definitely be more efficient than the way you currently have it.

CodePudding user response：

You can use np.is_busday

# Assuming date column is DatetimeIndex
hol = holidays.US(years=range(df['date'].dt.year.min(), df['date'].dt.year.max() 1))

df['business_day'] = np.is_busday(df['date'].astype(str).tolist(),
                                  weekmask='1111100', holidays=list(hol.keys()))

Output:

>>> df
        date  business_day
0 2011-07-03         False
1 2011-07-04         False
2 2011-07-05          True
3 2011-07-06          True
4 2011-07-07          True
5 2011-07-08          True
6 2011-07-09         False

CodePudding user response：

pip install swifter

and

import swifter
df.swifter.apply()