How to get the theoretic serial number of list elements if they were sorted by the sum in Python?-CodePudding

I have a list given filled with numpy arrays such as:

lst = [np.array([1, 2, 3, 4, 5]),
       np.array([16, 17, 18, 19, 20]),
       np.array([6, 7, 8, 9, 10]),
       np.array([11, 12, 13, 14, 15])]

I have already tried sort and sorted functions, but I did not find the proper key to monitor serial numbers.

I would like to find out which place would the elements (numpy arrays) of the list take if I sorted them based on their sum.

Considering the example, my expected output would be:

output = [0, 3, 1, 2]

Thank you very much in advance.

CodePudding user response：

Use np.argsort twice on the sum of elements:

import numpy as np

lst = [np.array([1, 2, 3, 4, 5]),
       np.array([16, 17, 18, 19, 20]),
       np.array([6, 7, 8, 9, 10]),
       np.array([11, 12, 13, 14, 15])]

print(np.argsort(np.argsort(np.sum(lst, 1))))

Output:

[0 3 1 2]

CodePudding user response：

You can use scipy.stats.rankdata:

import numpy as np

lst = [np.array([1, 2, 3, 4, 5]),
       np.array([16, 17, 18, 19, 20]),
       np.array([6, 7, 8, 9, 10]),
       np.array([11, 12, 13, 14, 15])]

from scipy.stats import rankdata
out = (rankdata(np.sum(lst, 1))-1).astype(int)

# array([0, 3, 1, 2])

CodePudding user response：

I think expected output should be [0,2,3,1]

You can use sorted

import numpy as np
list_ = [np.array([1, 2, 3, 4, 5]), np.array([16, 17, 18, 19, 20]), np.array([6, 7, 8, 9, 10]), np.array([11, 12, 13, 14, 15])]
print([i for i, _ in sorted(zip(range(len(list_)), list_), key=lambda x: sum(x[1]))])