how i select feature from four different weight technique
fatimidveil
New Altair Community Member
hi everyone hope you all are doing fine .
i am working on my thesis . i apply 4 feature selection techniques such as weight by information gain ,weight by information gain ratio ,weight by Gini index ,weight by Chi sqaure .
And Every technique rank, feature in different order .
what should i do now , how i figure out the best feature from these so that i have to apply my algorithm to find result .\
waiting for your reply
thanks
i am working on my thesis . i apply 4 feature selection techniques such as weight by information gain ,weight by information gain ratio ,weight by Gini index ,weight by Chi sqaure .
And Every technique rank, feature in different order .
what should i do now , how i figure out the best feature from these so that i have to apply my algorithm to find result .\
waiting for your reply
thanks
0
Best Answer
-
Hi @fatimidveil,
If you want to have average of them for each features, you can take a look at the sample process called "Predictive Maintenance"
and especially to the "step 2" of this process : The subprocess called "determine influence factors".
Hope this helps,
Regards,
Lionel
1
Answers
-
am using 4 operator(weight by Information Gain, weight by Information Gain Ratio, weight by Gini Index, weight by Chi Square). then i need to have average of them for each features.
0 -
Hi @fatimidveil,
If you want to have average of them for each features, you can take a look at the sample process called "Predictive Maintenance"
and especially to the "step 2" of this process : The subprocess called "determine influence factors".
Hope this helps,
Regards,
Lionel
1 -
thankyou fro your response and if i want to retrieve best feature from these 4 different criteria . is this possible ?
because individually information gain ratio has rank different feature then information gain gini index and chi sqaure
0 -
i want to collect best features from information gain information gain ratio gini index and chi sqaure collectively .
0 -
thank you Sir my problem is solved with your help thanks alot0