One of the most commonly used tools in public policy and social sciences are surveys. In the absence of directly observable actions, self-reported surveys become indispensable in identifying and classifying high-risk individuals. One such survey, which identifies individuals under financial duress, is the subject of our project. Using Support Vector Machines (SVM), a supervised machine learning method, we investigate whether indirect information from our survey dataset determines a respondent’s financial well-being. In addition, we utilize this dataset to investigate the dependencies of SVM. We find that we are able to correlate financial well-being with only 10% of the given questionnaire data, and that SVM performs best on two classes.