Obesity makes for good regression projects because it is correlated with many other items but causation is often dubious. For example, people who exercise tend not to be obese – right? But people who are obese find it hard to exercise, so the causation is not always clear. Similarly for diabetes, income level, and various health attributes.