Last active
November 25, 2015 12:37
-
-
Save thekensta/68e2cc4793fa8f847842 to your computer and use it in GitHub Desktop.
Numpy and Pandas date arithmetic
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import pandas as pd | |
import numpy as np | |
# | |
td = pd.DataFrame({'Date': pd.date_range('2014-01-01', '2015-12-31')}) | |
td['Timedelta'] = td['Date'].max() - td['Date'] | |
td.dtypes | |
# Date datetime64[ns] | |
# Timedelta timedelta64[ns] | |
# dtype: object | |
td["AgeDays"] = td["Timedelta"] / np.timedelta64(1, 'D') | |
td["AgeMonths"] = td["Timedelta"] / np.timedelta64(1, 'M') | |
td["AgeYears"] = td["Timedelta"] / np.timedelta64(1, 'Y') | |
td.head() | |
# Date Timedelta AgeDays AgeMonths AgeYears | |
# 0 2014-01-01 729 days 729 23.951210 1.995934 | |
# 1 2014-01-02 728 days 728 23.918356 1.993196 | |
# 2 2014-01-03 727 days 727 23.885501 1.990458 | |
# 3 2014-01-04 726 days 726 23.852646 1.987720 | |
# 4 2014-01-05 725 days 725 23.819791 1.984983 | |
# Days in a month | |
(td["AgeDays"] / td["AgeMonths"]).head() | |
# 0 30.436875 | |
# 1 30.436875 | |
# Days in a Year | |
(td["AgeDays"] / td["AgeYears"]).head() | |
# 0 365.2425 | |
# 1 365.2425 | |
## PERIODS | |
td["Month"] = td["Date"].dt.to_period("M") | |
td["Year"] = td["Date"].dt.to_period("Y") | |
td.head() | |
# Date Timedelta AgeDays Month Year | |
# 0 2014-01-01 729 days 729 2014-01 2014 | |
# 1 2014-01-02 728 days 728 2014-01 2014 | |
## TYPES | |
td.dtypes | |
# Date datetime64[ns] | |
# Timedelta timedelta64[ns] | |
# AgeDays float64 | |
# Month object | |
# Year object | |
# dtype: object | |
d0 = td["Date"].iloc[0] | |
p0 = td["Month"].iloc[0] | |
type(d0) | |
# pandas.tslib.Timestamp | |
type(p0) | |
# pandas._period.Period | |
# Work out the first date of the week or first date of the Month | |
week_start = pd.offsets.Week(weekday=0) | |
month_begin = pd.offsets.MonthBegin() | |
td["WeekStart"] = td.Date.apply(lambda x: week_start.rollback(x)) | |
td["MonthStart"] = td.Date.apply(lambda x: month_begin.rollback(x)) | |
print(td[["Date", "MonthStart", "WeekStart"]].head(25)) | |
# Date MonthStart WeekStart | |
# 0 2014-01-01 2014-01-01 2013-12-30 | |
# 1 2014-01-02 2014-01-01 2013-12-30 | |
# 2 2014-01-03 2014-01-01 2013-12-30 | |
# 3 2014-01-04 2014-01-01 2013-12-30 | |
# 4 2014-01-05 2014-01-01 2013-12-30 | |
# 5 2014-01-06 2014-01-01 2014-01-06 | |
# 6 2014-01-07 2014-01-01 2014-01-06 | |
# 7 2014-01-08 2014-01-01 2014-01-06 | |
# 8 2014-01-09 2014-01-01 2014-01-06 | |
# 9 2014-01-10 2014-01-01 2014-01-06 | |
# 10 2014-01-11 2014-01-01 2014-01-06 | |
# 11 2014-01-12 2014-01-01 2014-01-06 | |
# 12 2014-01-13 2014-01-01 2014-01-13 | |
# 13 2014-01-14 2014-01-01 2014-01-13 | |
# 14 2014-01-15 2014-01-01 2014-01-13 | |
# 15 2014-01-16 2014-01-01 2014-01-13 | |
# 16 2014-01-17 2014-01-01 2014-01-13 | |
# 17 2014-01-18 2014-01-01 2014-01-13 | |
# 18 2014-01-19 2014-01-01 2014-01-13 | |
# 19 2014-01-20 2014-01-01 2014-01-20 | |
# 20 2014-01-21 2014-01-01 2014-01-20 | |
# 21 2014-01-22 2014-01-01 2014-01-20 | |
# 22 2014-01-23 2014-01-01 2014-01-20 | |
# 23 2014-01-24 2014-01-01 2014-01-20 | |
# 24 2014-01-25 2014-01-01 2014-01-20 | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment