XGBoost regression

Question

I have a data set with events related to the customer behavior and the revenue the customers generated. I'd like to use XGBoost regression in Python to find which variables are more important in predicting the revenue. So from what I understand, I would use the feature importance to find that. Yet, I would also want to know how much changing an independent variable changes the dependent variable. Is there a way to do that, and if so pointers to any samples on that would be great. Any other ways to evaluate the impact of the independent variable using XGBoost besides the feature importance would be great.