Multi objective Evolutionary parameter optimisation
Contributor I
Hello,
I have used the "Optimize Selection (Evolutionary)" operator with multi-optimisation method "non-dominating" to minimise the number of attributes. This uses the "Performance (Attribute Count)" operator that provides the number of attributes along with the performance metric calculated by cross-validation, which enables the Evolutionary FS algorithm to build a pareto front using features v/s accuracy.
I am now trying to perform a similar multi-objective approach with the "Optimize Parameters (Evolutionary)" operator to optimise a Random Forest model while minimising a parameter such as the number of trees.
I'd be grateful if anyone could suggest a way to pass the parameter of an algorithm in my case - number of trees - as a performance metric within the "Optimize Parameters (Evolutionary)"?
Thanks in advance
Narayan
I have used the "Optimize Selection (Evolutionary)" operator with multi-optimisation method "non-dominating" to minimise the number of attributes. This uses the "Performance (Attribute Count)" operator that provides the number of attributes along with the performance metric calculated by cross-validation, which enables the Evolutionary FS algorithm to build a pareto front using features v/s accuracy.
I am now trying to perform a similar multi-objective approach with the "Optimize Parameters (Evolutionary)" operator to optimise a Random Forest model while minimising a parameter such as the number of trees.
I'd be grateful if anyone could suggest a way to pass the parameter of an algorithm in my case - number of trees - as a performance metric within the "Optimize Parameters (Evolutionary)"?
Thanks in advance
Narayan
0
