mRMR score and sorting

[saskra]

I am a bit confused why the mRMR table has one line more than the maxRel table - does it skip the least relevant feature in the first table, if you want to have ordered all of them?

And I do not understand the order in the second table - it does not seem to be ordered according to the actual score as you can see in this example.

*** mRMR features *** Order Fea Name Score 1 887 a64345 0.99573 2 889 a8BYI6 0.99990 3 670 a9WVK4 0.99990 4 837 b11688 0.99990 5 006 a9WVJ3 0.99990 6 434 a9CaD1 0.99925 7 420 a9DBG3 0.99754 8 888 b08071 0.99699 9 326 b19157 0.99677 10 892 a810a5 0.99571 11 893 c09046 0.99238 12 270 a8BGG7 0.99077 13 232 b31324 0.98236 14 117 a9CRB6 0.97790 15 882 b55097 0.97712 16 868 a9CYN9 0.97711 17 023 a61503 0.97775 18 785 a6bB93 0.97892 19 165 a922B1 0.97849 20 683 a9JIK5 0.97918

[saskra]

I have a result where the first entries in the MaxRel list have an identical score and the first one of these is taken as the first row in the MRMR table. Is the order in the MaxRel list for equal scores random and therefore the order in the second list as well? I have one feature among those I am especially interested in, can I force it to be the "first among equals" in the second table?

[bhpiamnothing]

I have the same problem. I don’t understand why in the {mRMR fearures} ，the first feature‘s score is lower than the second feature

[mainguyenanhvu]

Hi you,

Please show me the way to get score table of all features. I can't find the guideline for this.

Thank you very much. PS: The code I used:

pymrmr.mRMR(df, 'MIQ', 10)