Feature/Variable selection

hello there, I have a feature selection problem. I have a dataset with 14 features. I have done an exhaustive search of 2^P-1 to yield 16,383 subsets. I then clustered the generated subsets using k-means clustering algorithm. So I am looking to find the best subset. The information that I have is as follows (as an illustration).

Dataset cluster membership silhouette value Feature/Variables 1 3 1.00 1 2 4 1.00 2 3 4 0.97 3 4 4 0.80 4 5 3 0.79 5 6 3 0.64 [6,7] 7 2 0.37 [8,9,10,11,12,13,14] 8 4 0.48 [15,16] 9 2 0.66 [17,18,19,20,21,22,23,24] 10 4 1.00 25 11 3 0.38 [26,27,28,29,30,31,32,33,34] 12 3 0.77 35 13 3 0.79 36 14 2 0.76 37 15 3 0.78 [1,2] 16 3 0.94 [1,3] 17 3 0.93 [1,4] 18 2 0.73 [1,5] 19 2 0.64 [1,6,7] 20 4 0.39 [1,8,9,10,11,12,13,14] 21 2 0.62 [1,15,16] 22 3 0.50 [1,17,18,19,20,21,22,23,24] 23 4 0.67 [1,25] 24 2 0.53 [1,26,27,28,29,30,31,32,33,34] 25 4 0.86 [1,35] 26 3 0.86 [1,36]

any ideas would be appreciated. Thank you

Answer 1

[Amit Doshi]

Hello Eng,

You can use 'sequentialfs' function in the Statistics and Machine Learning Toolbox in MATALB to do sequential feature selection and reduce the dimensionality of data.

Refer the below links to know more about Feature Selection in MATLAB:

https://www.mathworks.com/help/stats/feature-selection.html