I have an experiment on the classification of probabilistic neural networks created in MATLAB. I can get classes for invisible data with the command sim. Is there a way to get the probabilities for the classes that the classifier computes? Also, is there any direct way to plot the receiver curve and calculate the area under ROC for my classifier?
sim
if you have a statistics panel, you can use the perfcurve function added in recent versions of MATLAB to build ROC curves and get AUC.
, , . ROC , PNN AUC - . , , -, , . , , . , stackoverflow , .
Source: https://habr.com/ru/post/1699518/More articles:Cancel request validation using HttpHandler in IIS 7 - c #QuickStart basic layout / template for web development? - templatesContains () and how to implement it - c #Add my linq query result set to the dataset, C # asp.net 3.5 - c #Is it possible to develop plugins for Microsoft Office Communicator? - ms-officeCan you use an XML-RPC data source for Adobe Flex applications? - flexFlex dataGrid itemRenderer, width and height of access cell - flexHow to set caching of static files in IIS? - iisWhen to change the GUID in a type library - guidAre styling tools useful? - coding-styleAll Articles