How to create a custom frequency in pandasper day at 9, 12 and 18 o'clock

Question

I understand how to create a pandas frequency like in python3:

import pandas as pd 

import datetime

idx = pd.date_range('2017-01-01' ,'2017-06-16',  freq='D')

ts = pd.Series(range(len(idx)), index=idx)

ts

How would I do this for irregularly sampled hourly data of 9 12 and 18 o'clock?

Answer 1

You can try:

idx = pd.date_range('2017-01-01' ,'2017-06-16',  freq='H')
idx = idx[idx.hour.isin([9,12,18])]

ts = pd.Series(range(len(idx)), index=idx)

Output:

2017-01-01 09:00:00      0
2017-01-01 12:00:00      1
2017-01-01 18:00:00      2
2017-01-02 09:00:00      3
2017-01-02 12:00:00      4
                      ... 
2017-06-14 12:00:00    493
2017-06-14 18:00:00    494
2017-06-15 09:00:00    495
2017-06-15 12:00:00    496
2017-06-15 18:00:00    497
Length: 498, dtype: int64