Returning an empty subset as the best subset for feature selection

[ahcantao]

• PySwarms version: 0.1.9
• Python version: 3.5.4
• Operating System: windows 10

Description

The code is returning an empty subset of attributes as the best subset selection. It is even returning the cost of the subset with only zeros. Even with a minor change on the exemple code it is possible to replicate this output.

There is a checking condition on the code to avoid empty subsets but it still outputs empty as the best subset.
if np.count_nonzero(m) == 0:
#if the particle subset is only zeros, get the original set of attributes
X_subset = X

Describe what you were trying to get done.
Return a non-empty subset of attributes (a binary list containing at least one element 1)

What I Did

I ran the "feature_subset_selection" notebook (https://github.com/ljvmiranda921/pyswarms/blob/master/examples/feature_subset_selection.ipynb) up to the # Perform optimization point. It all runs ok, but when I do a minor change, just on the number of iterations from 1000 to 10, it most of the times return an empty subset as the best subset.

The only line changed was the following:

cost, pos = optimizer.optimize(f, print_step=1, iters=10, verbose=2)
print(cost,pos)

and the output was:

INFO:pyswarms.discrete.binary:Iteration 1/10, cost: 0.2696
INFO:pyswarms.discrete.binary:Iteration 2/10, cost: 0.2696
INFO:pyswarms.discrete.binary:Iteration 3/10, cost: 0.2696
INFO:pyswarms.discrete.binary:Iteration 4/10, cost: 0.2696
INFO:pyswarms.discrete.binary:Iteration 5/10, cost: 0.2696
INFO:pyswarms.discrete.binary:Iteration 6/10, cost: 0.2696
INFO:pyswarms.discrete.binary:Iteration 7/10, cost: 0.2376
INFO:pyswarms.discrete.binary:Iteration 8/10, cost: 0.2376
INFO:pyswarms.discrete.binary:Iteration 9/10, cost: 0.2376
INFO:pyswarms.discrete.binary:Iteration 10/10, cost: 0.2376
INFO:pyswarms.discrete.binary:================================
Optimization finished!
Final cost: 0.2376
Best value: [ 0.000000 0.000000 0.000000 ...]

0.2376 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]

[ljvmiranda921]

Hi @ahcantao , sorry for the late reply.

Let me check on this real quick this weekend. Thanks for reporting!

[ahcantao]

@ljvmiranda921 Check if it is possible to avoid the creation of a subset containing only zeros. If this is possible, as a future update, you can add an extra parameter asking for the minimum quantity of 1's in the subset. Ex: minOnes = 3. Will return a subset containing 3 or more attributes out of the full set ([1 1 1 0 0 0 0 0 0 0].

[ljvmiranda921]

Thanks for your input, I'm currently in the process of updating PySwarms' backend (in development branch, not in master). I'd probably incorporate your idea of seeing a minimum number of attributes from the full set. Let's see where it goes.

[ljvmiranda921]

Note to self: make another PR to test this. This should be solved using the new backend implementation but further testing is needed.

[ljvmiranda921]

This should be solved by #145

[bhaskatripathi]

I am still facing the error for best subset for feature selection.

Perform optimization

cost, pos = optimizer.optimize(f,iters=700, fast=False)
The optimizer function returns the following output for pos-

1. I have 30 features and 903 rows
2. pos shows 903 returns an n-dimensional array of (903,30).
3. X_selected_features = X[:,pos==1] shows an "IndexError: too many indices for array" since it expects a one dimensional arrary something like this [0 0 0 0 1 1 0] instead, I am getting -
   array([[1, 1, 1, ..., 1, 1, 1],
   [1, 1, 1, ..., 1, 1, 1],
   [1, 1, 1, ..., 1, 1, 1],
   ...,
   [1, 1, 1, ..., 1, 1, 1],
   [1, 1, 1, ..., 1, 1, 1],
   [1, 1, 1, ..., 1, 1, 1]])