On May 11, 2005, at 8:49 PM, Praveen Boinee wrote:

> Dear WEKA People

>

> 1) How to plot multiple ROC 's on single plot ... I see nice graphics

> roc plot in WEKA for individual classifiers

> is thery any way to put all the ROC's in one plot

You can do that using the KnowledgeFlow interface.

> 2) I want to compare MLP with SVM and random forests

> my test set has 6400 rows ...i am using the explorer in WEKA

> I can see the False positive rate[FPR] and true pos.rate[TPR] for all

> 6400 rows only in MLP !! With SVM there only 2 rows of FPR and TPR

> and with random forests only 30 rows of FPR and TPR ..

>

> is it common to get like this ..!!

You need to use the -M option (i.e. fit logistic models) to get less

discrete "probability estimates" (the standard ones are based on votes,

i.e. 0/1 in the two-class case). Random forests produce pretty discrete

distributions because standard trees do. You might be able to get more

points by increasing the number of iterations.

Cheers,

Eibe

_______________________________________________

Wekalist mailing list

[hidden email]
https://list.scms.waikato.ac.nz/mailman/listinfo/wekalist