Open in new window / Try shogun cloud
--- Log opened Wed Jun 10 00:00:03 2015
shogun-buildbotbuild #1016 of nightly_default is complete: Failure [failed test notebooks]  Build details are at  blamelist: Heiko Strathmann <>, Wu Lin <>04:13
-!- zhuo [b45fe03b@gateway/web/freenode/ip.] has joined #shogun07:28
-!- zhuo [b45fe03b@gateway/web/freenode/ip.] has quit [Quit: Page closed]07:33
-!- ephas [89dea0c1@gateway/web/freenode/ip.] has quit [Quit: Page closed]09:06
-!- besser82 [~besser82@fedora/besser82] has joined #shogun09:41
-!- mode/#shogun [+o besser82] by ChanServ09:41
-!- HeikoS [] has joined #shogun10:01
-!- mode/#shogun [+o HeikoS] by ChanServ10:01
-!- HeikoS [] has quit [Quit: Leaving.]10:20
-!- shaochuan [] has joined #shogun10:50
-!- shaochuan [] has quit [Ping timeout: 264 seconds]10:55
-!- HeikoS [] has joined #shogun11:13
-!- mode/#shogun [+o HeikoS] by ChanServ11:13
CaBaare there any examples on the spectrum rbf kernels available anywhere?11:42
-!- HeikoS [] has quit [Quit: Leaving.]12:04
-!- HeikoS [] has joined #shogun13:08
-!- mode/#shogun [+o HeikoS] by ChanServ13:08
-!- HeikoS [] has quit [Ping timeout: 276 seconds]14:04
-!- HeikoS [] has joined #shogun15:53
-!- mode/#shogun [+o HeikoS] by ChanServ15:53
-!- HeikoS [] has quit [Client Quit]15:55
-!- ephas [89dea0c1@gateway/web/freenode/ip.] has joined #shogun16:03
ephasHi, if I use ROCEvaluation during cross validation, I am able to extract the mean AUC using ".mean".  I was wondering if there was a way to extract the tp/fp matrix you get from the get_ROC function, i.e. I am after the cross validated get_ROC matrix (I can only seem to extract the tp/fp matrix from the last fold and not the averaged matrix).  I hope this makes sense?16:08
ephasThe reason I am after this matrix is that I would like to draw an ROC curve from the cross validated results.16:09
ephasPS: I'm using the Python interface16:09
lisitsynephas: hey16:10
lisitsynyeah I got your problem16:10
lisitsynlet me check16:12
ephas@lisitsyn - thanks!16:12
lisitsynephas: you need averaged matrix right?16:13
ephasyes please16:14
lisitsynok this need some hack :(16:14
ephasrunning the .get_ROC function on the metric returns the matrix corresponding to the last fold16:14
ephasah - is the hack doable?16:15
lisitsyn.. in progress :)16:17
lisitsynlet me check again16:17
lisitsynephas: what about running it only once so that last fold is the only one?16:25
lisitsynand then again and again16:25
lisitsynthen manually compute mean ROC16:25
ephasI was thinking about that, but I couldn't work out how to calculate the mean matrix when the dimensions differ across folds (or is this consistent? ... I've not checked)16:28
lisitsynephas: I think they should match16:31
ephasok - I'll have a quick look and let you know how I get on16:32
ephasthanks for your help so far btw16:33
lisitsynnp, I'll try to help if any other issues occur16:34
-!- ephas [89dea0c1@gateway/web/freenode/ip.] has quit [Quit: Page closed]16:55
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 256 seconds]19:43
-!- HeikoS [] has joined #shogun20:09
-!- mode/#shogun [+o HeikoS] by ChanServ20:09
-!- HeikoS [] has quit [Quit: Leaving.]20:20
--- Log closed Thu Jun 11 00:00:04 2015