Building Machine Learning Models to Correct Self-Reported Information to Estimate Obesity Prevalence