Evaluation of my GBM ML model


I trained my model using GBM algo from Caret package. I fine tuned it using Grid search. Then I evaluated my model using confusion matrix and ROC value. Below is the numbers for evaluation -

  1. For Training Data - ROC - 0.94, Accuracy - 87.76%, Sensitivity- 88.67%, Specificity - 85.55%
  2. For Test Data - ROC - 0.94, Accuracy - 88.09%, Sensitivity- 87.64%, Specificity - 85.59%

I wanted to know that, whether my model is good or not, if not then what all parameters should I focus more for getting the generalized model?

"Good" is really defined by the problem that you are working on. An AUC of 0.94 would be envied by many people who model data though.