Having multi-class events, how do I assess the misclassification rate over all classes?

[acordes]

In Visual Analytics, whatever assessment statistiscs I choose, it refers to one specific level.

But how do I evaluate the model over all classes' misclassifiaction?

[PeterChristie]

Let me try to answer your question as I understand it. I ran a model to predict a nominal target with 9 levels. A specific level (A) is selected but the Misclassification rate for the overall model stays the same regardless. The bottom right graph is a Confusion Matrix showing observed and predicted results for all 9 levels. The table at the very bottom shows detailed numbers of observations and percentage for all possibilities.
You can click the Create Pipeline button to run further analysis in Model Studio and prepare the model for deployment. Model Studio gives you a nice Nominal Classification chart.

 

Model Studio Nominal Classification chart: