I was looking through a few scenarios where it might be okay to use variables that have high VIF here. But most of such discussions and remedies I have seen use it in the context of linear regression models.
So the question that I have is- Is multicollinearity a bad thing for all models. More specifically, do I need to check for multicollinearity when building the following models
- Random Forest
- Decision Trees
- Boosting methods (GBTrees and XGBoost)
Also, a second part I'd like to add is that- does it make a difference if I'm performing classification instead of regression. Do I need to interpret multicollinearity differently in these two cases?