Working with pd.Period is very slow

[benrifkind]

I find working with pd.Period very slow. Much much slower than working with pd.Timestamp and I was wondering why that is. I have around 2 million records and I'm forced to work with full timestamps even though my data is really on the level of months. Here is an example that I think shows the discrepancy between working with these types.

Code Sample, a copy-pastable example if possible

import pandas as pd
import numpy as np

# create a data frame with 500K dates
dates = pd.date_range('1/1/2011', periods=500000, freq='H')
values = np.random.random(size=len(dates))
df = pd.DataFrame({"Date":dates, "Values":values})

Creating the period index is fairly slow

%%time
df["Period"] = df.Date.dt.to_period(freq="1M")

Wall time: 1.47 s

And using boolean indexing with periods is extremely slow

%%time
near_future = df[df.Period <= df.Period.iloc[0] + 3] 

Wall time: 7.36 s

Especially when compared with using timestamps!

%%time
near_future = df[df.Date <= df.Date.iloc[0] + pd.Timedelta("90d")]

Wall time: 0 ns

Returning the values of the period is also very slow

%%time
periods = df.Period.unique()

Wall time: 17.4 s

%%time
times = df.Date.unique()

Wall time: 69 ms

Unrelated but I also found that I can only do the period comparison on a series in one direction. This throws an error even though all I did was flip the terms in the inequality

near_future = df[(df.Period.iloc[0] + 3 ) >= df.Period] 

---------------------------------------------------------------------------

TypeError                                 Traceback (most recent call last)

<ipython-input-8-f626f4b8a04f> in <module>()
----> 1 near_future = df[(df.Period.iloc[0] + 3 ) >= df.Period]


pandas\src\period.pyx in pandas._period.Period.__richcmp__ (pandas\src\period.c:12753)()


TypeError: Cannot compare type 'Period' with type 'Series'

Output of pd.show_versions()

INSTALLED VERSIONS
------------------
commit: None
python: 3.5.1.final.0
python-bits: 64
OS: Windows
OS-release: 7
machine: AMD64
processor: Intel64 Family 6 Model 61 Stepping 4, GenuineIntel
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8

pandas: 0.18.1
nose: 1.3.7
pip: 8.1.2
setuptools: 21.2.2
Cython: 0.24
numpy: 1.11.0
scipy: 0.17.0
statsmodels: 0.6.1
xarray: None
IPython: 4.1.2
sphinx: 1.3.1
patsy: 0.4.0
dateutil: 2.5.3
pytz: 2016.4
blosc: None
bottleneck: None
tables: 3.2.2
numexpr: 2.5.2
matplotlib: 1.5.1
openpyxl: 2.2.6
xlrd: 0.9.4
xlwt: 1.0.0
xlsxwriter: 0.7.7
lxml: 3.4.4
bs4: 4.4.1
html5lib: None
httplib2: None
apiclient: None
sqlalchemy: 1.0.11
pymysql: None
psycopg2: None
jinja2: 2.8
boto: 2.38.0
pandas_datareader: None

[jreback]

this is a symptom of #7964 it is an object dtype now (actually has been forever) and is not first class. contributions along those lines are welcome.