Create a machine learning model using logistic regression that can predict credit card approvals from the described dataset.
The linear model is then expanded using polynomial regression and the accuracies obtained are examined using confusion matrices and precision-recall curves.