How to sort a 2D array of two columns, from large to small, by the second column?-CodePudding

Here is my code:

unique_models, count_of_models = np.unique(my_data_frame["model"], return_counts=True)
print(unique_models, count_of_models)   
                                        
[' A1' ' A2' ' A3' ' A4' ' A5' ' A6' ' A7' ' A8' ' Q2' ' Q3' ' Q5' ' Q7' ' Q8' ' R8'                         
 ' RS3' ' RS4' ' RS5' ' RS6' ' RS7' ' S3' ' S4' ' S5' ' S8' ' SQ5' ' SQ7' ' TT'] 
[1347    1 1929 1381  882  748  122  118  822 1417  877  397   69   28   33   31   29   
 39    1   18   12    3    4   16    8  336]

representative_models = np.empty((0, 2), int) 
other_models = np.empty((0, 2), int)

for models, counts in zip(unique_models, count_of_models):

    if counts < 500:

        other_models = np.append(other_models, np.array([[models, counts]]), axis=0)

    else:

        representative_models = np.append(representative_models, np.array([[models, counts]]), axis=0)

print(representative_models[representative_models[:, 1].argsort()])

[[' A1' '1347']
 [' A4' '1381']
 [' Q3' '1417']
 [' A3' '1929']
 [' A6' '748']
 [' Q2' '822']
 [' Q5' '877']
 [' A5' '882']]  

print(representative_models)

[[' A1' '1347']
 [' A3' '1929']
 [' A4' '1381']
 [' A5' '882']
 [' A6' '748']
 [' Q2' '822']
 [' Q3' '1417']
 [' Q5' '877']]

So as you can see, everything was succesful except for the sorting, they are simply not sorted. Anyone would know of another method to sort from largest to smallest by the second column???

Example of what it should look like:

[[' A3' '1929']
 [' Q3' '1417']
 [' A4' '1381']
 [' A1' '1347']
 [' A5' '882']
 [' Q5' '877']
 [' Q2' '822']
 [' A6' '748']]

Thank you!

CodePudding user response：

Here you go:

import numpy as np
data = [[' A1', '1347'],
 [' A4', '1381'],
 [' Q3', '1417'],
 [' A3', '1929'],
 [' A6', '748'],
 [' Q2', '822'],
 [' Q5', '877'],
 [' A5', '882']] 
indices = np.argsort([int(d[1]) for d in data])
sorted_data = [data[i] for i in indices[::-1]]

CodePudding user response：

Assuming your array is as below, you could do:

import numpy as np

representative_models = np.array([[' A1', '1347'],
                                  [' A4', '1381'],
                                  [' Q3', '1417'],
                                  [' A3', '1929'],
                                  [' A6', '748'],
                                  [' Q2', '822'],
                                  [' Q5', '877'],
                                  [' A5', '882']])

# convert last column to int and arg-sort in decreasing order [::-1] 
order = np.argsort(representative_models[:, 1].astype(int))[::-1]

# simply index on the input array
result = representative_models[order, :]

print(result)

Output

[[' A3' '1929']
 [' Q3' '1417']
 [' A4' '1381']
 [' A1' '1347']
 [' A5' '882']
 [' Q5' '877']
 [' Q2' '822']
 [' A6' '748']]