Python3: Split Time Series By Diurnal Periods

I have the following dataset: 01/05/2020,00,26.3,27.5,26.3,80,81,73,22.5,22.7,22.0,993.7,993.7,993.0,0.0,178,1.2,-3.53,0.0 01/05/2020,01,26.1,26.8,26.1,79,80,75,22.2,22.4,21.9,994.

Solution 1:

You can use pd.cut:

bins = [-1,5,11,17,24]
labels = ['morning', 'afternoon', 'evening', 'night']

df['day_part'] = pd.cut(df['hour'], bins=bins, labels=labels)

Solution 2:

I added column names, including Hour for the second column.

Then I used read_csv which reads the source text, "dropping" leading zeroes, so that Hour column is just int.

To split rows (add a column marking the diurnal period), use:

df['period'] = pd.cut(df.Hour, bins=[0, 6, 12, 18, 24], right=False,
    labels=['night', 'morning', 'afternoon', 'evening'])

Then you can e.g. use groupby to process your groups.

Because I used right=False parameter, the bins are closed on the left side, thus bin limits are more natural (no need for -1 as an hour).

And bin limits (except for the last) are just starting hours of each period - quite natural notation.