How the housing industry is preparing for the age of AI

Machine Learning

Artificial intelligence (AI) and machine learning (ML) offer a fantastic opportunity to expand home ownership and housing in the United States. We need to find ways to remove bias from decision-making models, he said Monday.

“In my opinion, the greatest short- to medium-term opportunity remains. [finding] It’s a way to take unstructured data, make use of it, and turn it into machine-readable machine information. ” fannie mae.

Holden spoke on the topic of using data science and alternative data to increase access to home ownership. Mortgage Bankers AssociationThe (MBA’s) Technology Solutions Conference & Expo will be held Monday in San Jose, CA.

Fannie Mae’s focus on embedding machine learning is to obtain a rich profile of consumer risk and use it to assess their suitability for homeownership success.

In particular, Holden looked at the opportunity to train a computer algorithm to check bank statements and identify borrowers with poor credit but paying a fixed monthly rent.

“(Consistent rent payments) should be considered in the (renter’s) risk assessment (…) If you are a renter, these payments are not considered. This idea of ​​being able to extract that data and leverage it from decision-making is a very powerful and important innovation for accessing specific sectors of the population,” Holden said.

rocket company We focus on using technology to leverage our potential customer base.

Brian Stucky, Rocket Ethical AI Lead rocket central, noted the rapidly increasing rate of Hispanic homeownership. By 2040, 70% of new homeowners will be Latino, many of whom are at risk of being denied a mortgage due to high levels of debt to income.

“If Hispanics are dropping out because of DTI, what does that mean? What’s the part? In some way, the inclusion of those aspects that could help us identify clients who are currently turned down but are likely to be a good credit risk to us,” says Stuckey. said.

From a housing supply perspective, machine learning models can help address biases in the collateral valuation process. core logicSaid.

“I have seen the entire construction project in its early stages (…) there is no mortgage financing due to the perception that there will be no properties to sell in the neighborhood. , so there is no point in doing the project,” says Carroll.

The United States is facing a housing shortage of 1.1 million units, with an increase in entry-level homes ranging from 1 to 4 units of single-family homes, according to the U.S. Government. core logic.

Carroll said there was tremendous appetite from states, mortgage lenders, city and county local subsidy departments, and zoning departments to participate in the conversation, highlighting technology opportunities in addressing the housing shortage. I mentioned

The housing industry needs to figure out how to incorporate artificial intelligence and machine learning while preventing biases from creeping into business decision-making models, he said.

“You may have heard machine learning referred to as a black box, but it is difficult to see and understand why decisions are made from given inputs. We have to overcome this before we can adopt and use it,” Stucky explains.

Holden says that in a highly regulated industry like housing, it will be important to explain the reasons behind the decisions taken from the model.

“One of the things we often think about is explaining what those results are, what they mean and how they are generated when making decisions from models or doing analysis from models. When you start relying heavily on machine learning and AI-type methodologies, it becomes much more difficult,” said Holden.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *