Prediction plot improvement

[tortorellas]

Prediction plot is poorly interpretable;

1. I cannot resize the plot to properly inspect the results
   

2. I am trying to project on a PLS model some of the same data used to build the model itself…so I wasn’t expecting so poor predictive ability: any chance the algorithm is corrupted?

Version 2.3.0 (Windows), libscientific 1.4.1

[gmrandazzo]

Thank you! Yes! Ugly. I know. I'm working on a new plotting solution that is fast and better looking.
Anyway, looking at your plot, the object has some features out of the latent space.
That's why it is far away from the object space. Is it possible to investigate this point there?
To do that, please click on the point and show the label. Then click on "Data" -> "table where come from the point" -> right button of the mouse and then "Show data". Then look for the object name and check feature by feature carefully.
Do you spot anything "strange" in your numbers? Do you have in one of the features a value 1+e8?