[14:51:02] hi channel [14:51:27] o/ [14:57:30] ciao ciao! [18:12:07] lzia: one question if you may [18:12:28] here for you [18:12:31] nuria: [18:13:01] leila: I am now doubting one thing (about bots), the rates we talked about calculating were false positive rates [18:13:24] leila: that is , "positive migdiagnosis" [18:13:24] nuria: which rates? :) [18:13:32] leila: sorry sorry [18:13:45] leila: in the context of the bot work [18:14:03] leila: and the calculations of variables such us precision and recall for the classifier [18:14:09] leila: makes sense thus far? [18:14:13] yup [18:14:35] leila: we talked about calculating the false positive rate plus precion and recall correct? [18:15:02] * leila thinks. 1 sec [18:15:33] leila: doc in question https://docs.google.com/document/d/1q14GH7LklhMvDh0jwGaFD4eXvtQ5tLDmw3UeFTmb3KM/edit# (for reference, you do not have to read it now) [18:15:47] nuria: yes. we talked about calculating the confusion matrix. [18:16:21] (ow wow. this is turning to a doc.) [18:16:45] leila: ok, i can calculate confusion matrix with absiolute numbers rather than rates [18:17:11] * leila checks the definitions again [18:17:15] leila: you are an author on this doc NEWFLASH [18:17:30] yup. just noticed. ;) [18:18:56] nuria: if you compute the confusion matrix, then all those other rates and terms we can compute based on. [18:19:17] leila: ok, will do that if that is the preference , many thanks [18:19:24] np. [18:19:46] nuria: while you're at it, you may want to consider submitting it to Wiki Workshop http://wikiworkshop.org/2019/#call [18:19:53] the feedback can be helpful. [18:20:05] nuria: we can talk about it later if you're interested. [18:20:27] leila: it is a bit raw don't you think? but sure if you feel there is enough i can do it [18:20:55] nuria: you can do a short paper. let's talk on Friday? [18:20:56] leila: joseph has alredy looked at it and i need to incorportae his feedback , i also need to share with team wich i will do today [18:21:00] leila: k [18:21:07] sounds good [18:21:11] leila: many thanks, will set up a short sync up [18:21:35] sounds good. (i did NOT expect a bot question on Monday. I'm proud of you.) [19:30:53] leila: I found an old task re: sampling by log-in status in QuickSurvey and added it to the Research board, and you as a subscriber: https://phabricator.wikimedia.org/T186737 [21:28:01] J-Mo: thanks re the phab ticket and the comment there. [21:30:49] np leila. Hope it helps us lobby to improve the tool