What is the bias-variance tradeoff, and why is it important in predictive analytics? How do bias and variance individually affect model performance and generalization? What happens when a model has high bias or high variance? How can the bias-variance tradeoff be balanced to achieve optimal model performance? What techniques are used to manage bias and variance in predictive models?