CODEWITHSUNDEEP
pythondatetimepandas
Rich Signell
Rich Signell

Reputation: 16385

Pandas datetime column to ordinal

I'm trying to create a new Pandas dataframe column with ordinal day from a datetime column:

import pandas as pd
from datetime import datetime

print df.ix[0:5]
                              date
file                              
gom3_197801.nc 2011-02-16 00:00:00
gom3_197802.nc 2011-02-16 00:00:00
gom3_197803.nc 2011-02-15 00:00:00
gom3_197804.nc 2011-02-17 00:00:00
gom3_197805.nc 2011-11-14 00:00:00

df['date'][0].toordinal()

Out[6]:
734184

df['date'].toordinal()

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-7-dbfd5e8b60f0> in <module>()
----> 1 df['date'].toordinal()

AttributeError: 'Series' object has no attribute 'toordinal'

I guess this is a basic question, but I've struggled reading docs for last 30 minutes.

How can I create an ordinal time column for my dataframe?

Upvotes: 17

Views: 22910

Answers (4)

russhoppa
russhoppa

Reputation: 403

A 2x faster approach

I hate to have to resort to apply or map so here's a more efficient approach (about 2x faster in my case). It uses np.vectorize.

import pandas as pd
import numpy as np

def to_ordinal(dt):
  return dt.toordinal()

vectorized_ordinal = np.vectorize(to_ordinal, otypes=['int'])
df = pd.DataFrame()
df['date'] = pd.date_range('2000-01-01', '2030-01-01', freq='d')
df['ordinal_date'] = vectorized_ordinal(dates)

Using np.vectorize

%timeit vectorized_ordinal(df['date'])

5.89 ms ± 447 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

Using apply

%timeit df['date'].apply(pd.Timestamp.toordinal)

11.2 ms ± 429 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)

Using map

%timeit df['date'].map(pd.Timestamp.toordinal)

32.5 ms ± 1.74 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)

Upvotes: 1

wwii
wwii

Reputation: 23773

For completeness:
Apply pd.Timestamp.toordinal

df['date'].apply(pd.Timestamp.toordinal)

Upvotes: 3

behzad.nouri
behzad.nouri

Reputation: 77991

you can also use map:

import datetime as dt
df['date'].map(dt.datetime.toordinal)

Upvotes: 12

lowtech
lowtech

Reputation: 2582

Use apply:

df['date'].apply(lambda x: x.toordinal())

Upvotes: 29

Related Questions