High blood pressure is a leading cause of cardiovascular disease, with billions of dollars spent in the U.S. every year to treat hypertension. Hypertension has even been classified as an epidemic by some researchers and doctors. As such, it is important to understand who is at risk of developing high blood pressure, so that, hopefully, preventative steps can be taken. Using data acquired from the National Health and Nutrition Examination Survey (NHANES), we developed a regression model to predict systolic blood pressure from demographics and consumer data such as age, gender, education level, marital status, money spent on groceries and restaurants and how far away participants live from a grocery store. We converted the categorical variables we used – income, marital status, education – into dummy variables so that data could be more accurately analyzed. The data were then transformed by using the natural log in order to make it as normally distributed as possible. Variance inflation factors (VIF) were used to see if multicollinearity existed among any of the variables. Since all of the variables had VIF values less than ten, we concluded that the variables were not correlated and proceeded with the data analysis. In our preliminary regression models, we found that all of the demographics variables we used contributed significantly to blood pressure, but only one consumer behavior variable, distance from nearest grocery store, did. We will also find a parsimonious model by choosing the more parsimonious of two competing models which have essentially the same predictive power.
