Often, the first few iterations of some Visual ML algorithms seem to consistently produce R2 scores on the order of -10^25. Within just a few iterations, these usually jump up to a range that's meaningful. Unfortunately, when these numbers are plotted on the session chart, it becomes impossible to track performance over time without hovering over each point. It'd be great if the scale of this chart could be limited to some maximum so that a typical curve can be viewed consistently.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.