How to access Object attributes within numpy array structure

Question

Given a numpy array structure of identical (user specified) objects, is there a way to references all of them at once?

E.g. given a numpy array structure of objects of type date, is there a way to take the average of the years without resorting to a for loop or to +1 to the year attribute of each object in the array?

Example code follows.

from numpy import *
from datetime import *

#this works
A = array([2012, 2011, 2009])
print average(A)

date1 = date(2012,06,30)
date2 = date(2011,06,30)
date3 = date(2010,06,30)
B = array([date1, date2, date3])
print B[0].year
print B[1].year
print B[2].year

#this doesn't
print average(B.year)

Artsiom Rudzenka · Accepted Answer · 2011-08-18 11:11:01Z

1

Think you can do this the following way:

from numpy import array, average
from datetime import date

date1 = date(2012,06,30)
date2 = date(2011,06,30)
date3 = date(2010,06,30)
B = array([date1, date2, date3])

avYear = average([x.year for x in B])

EDITED as per comment:

B = array([x.replace(year=x.year+10) for x in B])

And note that using from module import * is not very good - it is always better to import only thoose classes and functions which you really need.

edited Aug 18, 2011 at 11:11

answered Aug 16, 2011 at 12:40

Artsiom Rudzenka

29.3k5 gold badges36 silver badges53 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

rm. Over a year ago

Thanks for your quick reply! I would appreciate it very much if you could also consider the following. How to subtract 10 years from each of the year attributes (2012,2011,2010) and store the new values in the same objects (of the structure B)?

Artsiom Rudzenka Over a year ago

@rm check my addition within answer whether it helps you or not

rm. Over a year ago

Thanks for the update! Could you please clarify one detail? Does the line “[x.replace(year=x.year+10) for x in B]” create new copies of the date objects or does it manipulate the original ones? After some testing it seems to me that it creates a new list with copies of the original date objects.

Artsiom Rudzenka Over a year ago

@rm as far as i know date object is not mutable, when i try to assign new value to year property it throwing an exception about read-only property editing. However think that you can update your code to store date values as a strings and handle(convert to date) them when needed.

Yaroslav Nikitenko · Accepted Answer · 2017-06-10 13:36:25Z

1

This can be done via vectorize function.

import numpy as np
from datetime import date

date1 = date(2012,06,30)
date2 = date(2011,06,30)
date3 = date(2010,06,30)
B = np.array([date1, date2, date3])

yr = lambda x: x.year
vyr = np.vectorize(yr)
print vyr(B)
# array([2012, 2011, 2010])
print np.average(vyr(B))
# 2011.0

Note from the manual:

The vectorize function is provided primarily for convenience, not for performance. The implementation is essentially a for loop.

answered Jun 10, 2017 at 13:36

Yaroslav Nikitenko

1,8732 gold badges25 silver badges32 bronze badges

Collectives™ on Stack Overflow

How to access Object attributes within numpy array structure

2 Answers 2

4 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

4 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related