What is a good RMSE value?

Hi there,

In this case I would recommend a residual analysis which I explained in this post here: How to evaluate accuracy of a regression model - #21 by Christian_Simonis

So if your model would be really underfitting but based on you residual analysis you would find out:

  • no patterns or correlation between features and residuum
  • from business perspective the residuals (represented with a histogram or so) are sufficiently good with respect to a performance metric (e.g. a certain quantile or worse case error)

then you can conclude that your model is “acceptable”.

But there are ways to do better of course: on this note you can check this thread to get some inspiration on how to tackle underfitting: Training set error? - #2 by Christian_Simonis

Hope that helps, @mehmet_baki_deniz.

Best regards

1 Like