Homework answers / question archive / In this week's discussion, we will examine the factors that contribute to diabetes

In this week's discussion, we will examine the factors that contribute to diabetes

Unrecognized

Share With

In this week's discussion, we will examine the factors that contribute to diabetes. Our data come from Efron, B., Hastie, T., Johnstone, J., and Tibshirani, R. (2004). Least Angle Regression. Annals of Statistics, 32, 407-499. These researchers use ten baseline variables (age, gender, body mass index, average blood pressure, and six blood serum measurements) for each of n = 442 diabetes patients. The response Y is a quantitative measure of disease progression one year after baseline. The response Y Binary is a nominal variable, indicating whether the disease has gotten better or worse (low, high). It is constructed from the response Y and is defined by its column formula. You could examine the disease progression both as a regression problem (by focusing on Y) or as a classification problem (by focusing on Y Binary). For this assignment, we will look at it as a classification problem. Data set is below, and also, if you click on the columns you will see the medical definitions of variables. The methods you will consider in this assignment are ordinary logistic regression and penalized logistic regression (lasso, adaptive lasso, elastic net, adaptive elastic net). Follow our course methodology: - create a cross validation column (note this is cross sectional data, so use JMP's make validation column facility and set random seed to 123) - estimate candidate models and save predicted probabilities - model comparison and choose best (by AUC) - interpret model parameters and variable importance Finally, as a new case, what is the predicted probability of progression to High for a patient with the following numbers: Age: 47, Gender=1, BMI=45, BP=109, Total Cholesterol=237, LDL=100.2, HDL=70, TCH=3, LTG=5.2149, Glucose=107