I have a Pandas dataframe with a DataTimeIndex and some other columns, similar to this:

import pandas as pd
import numpy as nprange = pd.date_range('2017-12-01', '2018-01-05', freq='6H')
df = pd.DataFrame(index = range)# Average speed in miles per hour
df['value'] = np.random.randint(low=0, high=60, size=len(df.index))df.info()
# DatetimeIndex: 141 entries, 2017-12-01 00:00:00 to 2018-01-05 00:00:00
# Freq: 6H
# Data columns (total 1 columns):
# value    141 non-null int64
# dtypes: int64(1)
# memory usage: 2.2 KBdf.head(10)
#                      value
# 2017-12-01 00:00:00     15
# 2017-12-01 06:00:00     54
# 2017-12-01 12:00:00     19
# 2017-12-01 18:00:00     13
# 2017-12-02 00:00:00     35
# 2017-12-02 06:00:00     31
# 2017-12-02 12:00:00     58
# 2017-12-02 18:00:00      6
# 2017-12-03 00:00:00      8
# 2017-12-03 06:00:00     30

How can I select or filter the entries that are:

1. Weekdays only (that is, not weekend days Saturday or Sunday)

2. Not within N days of the dates in a list (e.g. U.S. holidays like '12-25' or '01-01')?

I was hoping for something like:

df = exclude_Sat_and_Sun(df)omit_days = ['12-25', '01-01']
N = 3 # days near the holidays
df = exclude_days_near_omit_days(N, omit_days)

I was thinking of creating a new column to break out the month and day and then comparing them to the criteria for 1 and 2 above. However, I was hoping for something more Pythonic using the DateTimeIndex.

Thanks for any help.

The first part can be easily accomplished using the Pandas DatetimeIndex.dayofweek property, which starts counting weekdays with Monday as 0 and ending with Sunday as 6.

df[df.index.dayofweek < 5] will give you only the weekdays.

For the second part you can use the datetime module. Below I will give an example for only one date, namely 2017-12-25. You can easily generalize it to a list of dates, for example by defining a helper function.

from datetime import datetime, timedeltaN = 3df[abs(df.index.date - datetime.strptime("2017-12-25", '%Y-%m-%d').date()) > timedelta(N)]

This will give all dates that are more than N=3 days away from 2017-12-25. That is, it will exclude an interval of 7 days from 2017-12-22 to 2017-12-28.

Lastly, you can combine the two criteria using the & operator, as you probably know.

df[(df.index.dayofweek < 5) & (abs(df.index.date - datetime.strptime("2017-12-25", '%Y-%m-%d').date()) > timedelta(N))]