2

I have created 2 variables. One to hold 200 randomly generated ages, the other to hold 200 randomly generated marks.

from numpy import *
age = random.random_integers(18,40, size=(1,200))
marks = random.random_integers(0,100, size=(1,200))

I'd like to use NumPy to sort the marks array by the age array. eg:

#random student ages
[32 37 53 48 39 44 33 40 56 47]
#random student marks
[167 160 176 163 209 178 201 164 190 156]

#sorted marked according to ages
[32 33 37 39 40 44 47 48 53 56]
[167 201 160 209 164 178 156 163 176 190]

It is a similar question to this one. I am just unsure if a similar solution applies due to the elements being randomly generated.

Hamza Ebrahim
  • 27
  • 5
  • 1
    yes your linked answer does apply to your situation.. in your case you would sort `marks` according to `age` with: `sorted_marks = marks[age.argsort()]`. For this to work smoothly however you should eliminate the extraneous axis in your array by using `size=200` instead of `size=(1,200)` – Aaron May 29 '18 at 13:35
  • 2
    Possible duplicate of [How can I "zip sort" parallel numpy arrays?](https://stackoverflow.com/questions/1903462/how-can-i-zip-sort-parallel-numpy-arrays) – Aaron May 29 '18 at 13:39
  • better [example](https://stackoverflow.com/questions/9007877/sort-arrays-rows-by-another-array-in-python) – Aaron May 29 '18 at 13:42

1 Answers1

1

One way is to first calculate an ordering via argsort, then use this to index your input arrays::

import numpy as np

np.random.seed(0)

ages = np.random.randint(18, 40, size=10)   # [30 33 39 18 21 21 25 27 37 39]
marks = np.random.randint(0, 100, size=10)  # [36 87 70 88 88 12 58 65 39 87]

order = ages.argsort()                      # [3 4 5 6 7 0 1 8 2 9]

print(ages[order])                          # [18 21 21 25 27 30 33 37 39 39]
print(marks[order])                         # [88 88 12 58 65 36 87 39 70 87]
jpp
  • 159,742
  • 34
  • 281
  • 339