We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
ANALYZING RISKS FOR HOME LOANS DURING FINANCIAL CRISIS OF 2008 USING DATA MINING.
- Authors
Kim, Ben; Tunas, Nina; Green, Adrian
- Abstract
Freddie Mac's single family loan data provide ample information on historical single family loan originations and performances throughout the loan periods. We analyzed these datasets to answer a critical question for the secondary mortgage market: Were there identifiable characteristics or factors surrounding borrowing and lending leading up to the financial crisis that could have signaled early risks in the resulting mortgage backed securities? For this analysis, the data randomly sampled by Freddie Mac, including the origination and servicing data from year 2002 and 2007, were used. Each year's dataset consists of about a million unique loans and their performance data. After an iterative process of data cleaning and developing several models, we applied the Decision Tree algorithm to identify the major loan risk factors. Initially we thought credit scores might be a major factor for predicting risks. However, we found that MSA (Metropolitan Statistical Area), loan purposes, and original UPB (Unpaid Principal Balance) are the three most significant factors.
- Subjects
FREDDIE Mac (Company); FAMILY loans; FINANCIAL risk
- Publication
Issues in Information Systems, 2016, Vol 17, Issue 1, p206
- ISSN
1529-7314
- Publication type
Article