How does Entropy measure uncertainty in a dataset during machine learning classification? What role does information gain play in selecting decision tree splits? How can changes in entropy affect the structure and performance of a decision tree model?