Ranking Data Based on Other Results

user_13891514 0
Viewed 41

I have two sets of data in the below code that are linked - 5 'networks', and 5 singular results through 'TT' values. Each TT value corresponds to a network set (e.g. TT1=100 is a result from network1).

My goal is to be able to rank the 5 networks based on the TT results.

network1 = [0,1,0,0,0,0,1,0,0,0,1,1,0,0,1]
network2 = [0,1,0,0,0,0,0,0,1,0,1,1,0,0,1]
network3 = [0,1,0,0,0,0,0,1,0,0,1,1,0,0,1]
network4 = [0,1,0,0,0,0,0,0,0,0,1,1,0,1,1]
network5 = [0,1,1,0,0,0,0,0,0,0,1,1,0,0,1]

networks = [network1, network2, network3, network4, network5]

TT1 = 100
TT2 = 70
TT3 = 80
TT4 = 105
TT5 = 120

TTranks = [TT1,TT2,TT3,TT4,TT5]

s = {x: i for i, x in enumerate(sorted(set(TTranks)))}
ranks = [s[x] for x in TTranks] 
print(ranks)

The results below from this process rank the TT values only, although I am unsure how to link it so that the corresponding networks are also ranked appropriately.

[2, 0, 1, 3, 4]

That is, I would like it to then rank the 5 networks as network2, network3, network1, network4, network5. This will be very useful as I would like to be able to call from the ordered network results later on.

Additionally, is there a more effective way of completing this process?

I appreciate all the help!!

ranking sorting python
2 Answers

You can use the key optional argument of the sort/sorted functions to sort according to a custom value.

Code

network1 = [0,1,0,0,0,0,1,0,0,0,1,1,0,0,1]
network2 = [0,1,0,0,0,0,0,0,1,0,1,1,0,0,1]
network3 = [0,1,0,0,0,0,0,1,0,0,1,1,0,0,1]
network4 = [0,1,0,0,0,0,0,0,0,0,1,1,0,1,1]
network5 = [0,1,1,0,0,0,0,0,0,0,1,1,0,0,1]

networks = [network1, network2, network3, network4, network5]

TT1 = 100
TT2 = 70
TT3 = 80
TT4 = 105
TT5 = 120

TT = [TT1, TT2, TT3, TT4, TT5]

sorted_networks_with_indices = sorted(enumerate(networks), key=lambda t: TT[t[0]])

sorted_indices = [i for i,n in sorted_networks_with_indices]
sorted_networks = [n for i,n in sorted_networks_with_indices]

print(sorted_indices)
print(sorted_networks)

Output:

[1, 2, 0, 3, 4]
[[0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 1],
 [0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 1, 1, 0, 0, 1],
 [0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 1],
 [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 1, 1],
 [0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1]]

Code explanation

enumerate(networks) lists the pairs i,n where i is the index and n the network. These pairs are sorted with the sorted function, and the key used to sort is lambda t: TT[t[0]]. Here t is the pair i,n, so t[0] is the index i. Therefore TT[t[0]] is the TT corresponding to index-network pair t.

Note about 0-index

The indices in sorted_indices are [1, 2, 0, 3, 4] and not [2, 3, 1, 4, 5] because python lists are 0-indexed, but this corresponds to network2, network3, network1, network4, network5. If you are not happy with that you can replace the relevant line with:

sorted_indices = [i+1 for i,n in sorted_networks_with_indices]

References

user_30807230

Could you please explain how networks and TTranks are connected? The code shows no dependency on these two.

You simply want the indices for sorting TTranks? Then you should avoid set(TTranks), since the order will not be preserved in a set, also duplicates will be deleted. There are several ways to get the sorting indices, I prefer argsort from numpy:

import numpy as np
print(np.argsort(TTranks)) 
# [1 2 0 3 4]
user_80806480
Related