{"667348":{"#nid":"667348","#data":{"type":"news","title":"Coaching Tool Guides Rejected Loan Applicants Toward Better Outcomes","body":[{"value":"\u003Cp\u003EA new web-based tool is set to provide people with unprecedented visibility into the machine learning models that are used to make high-stakes decisions impacting their daily lives.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDeveloped at Georgia Tech, GAM Coach is the first interactive tool of its kind to give people with rejected loan applications the power to personalize recourse options that are realistically actionable to help ensure a better outcome in the future.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EExisting machine learning (ML) models generate recourse options based on fixed assumptions about a broad spectrum of people. GAM Coach, however, lets users iteratively adjust loan application features, such as loan amount, payment terms, credit score, homeownership status, and more, based on their personal preferences. \u0026nbsp;\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u201cWe can\u2019t assume that developers can make the best decisions for everyone,\u201d said Zijie (Jay) Wang, lead researcher and a Ph.D. student in Georgia Tech\u2019s School of Computational Science and Engineering (CSE).\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u201cOur goal is to give agency to the end user, so we developed GAM Coach to give people actionable recourse in scenarios like loan applications.\u201d\u003C\/p\u003E\r\n\r\n\u003Cp\u003EGAM Coach lets users exercise this agency by developing up to five recourse plans at a time. They can customize each iteration by adjusting sliders to set acceptable ranges for loan amount, revolving balance, and similar variable features. Emojis with related text like, \u201c\u2639\u0026nbsp;Very hard to change\u201d, are used to set difficulty levels for features that might be easier or harder to change depending on the individual.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EThe open-source tool is designed to let users \u201cplay around,\u201d says Wang, to see how adjusting one feature can impact the model\u2019s prediction. \u2019What if I raise my FICO score 10 points?\u2019 \u2018What if I reduce the loan amount?\u2019 \u2018What if I had 10% less debt?\u2019\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u201cNot every option is actionable for every person, but by allowing users to interact directly with their variable preferences, GAM Coach can find the minimal number of changes an individual needs to increase the likelihood of being approved for a loan,\u201d said Wang, a\u0026nbsp;\u003Ca href=\u0022https:\/\/www.cc.gatech.edu\/news\/student-named-apple-scholar-connecting-people-machine-learning\u0022\u003Erecipient of the 2023 Apple Scholars in AI\/ML PhD fellowship\u003C\/a\u003E.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EIf the initial five plans aren\u2019t satisfactory, users can continue to iteratively fine-tune their recourse options until they find a plan that best meets their needs.\u003C\/p\u003E\r\n\r\n\u003Cp\u003ETo build a tool that can generate personalized recourse options that are realistically actionable, Wang and his collaborators first developed an innovative new linear integer algorithm and an easy-to-use interactive data visualization interface. These were paired and then put under the hood of a generalized additive model (GAM) to create GAM Coach.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EGAMs are relatively common predictive ML models that are well-suited for determining optimal solutions. They\u2019re also known for their simplicity and transparency, which is a big reason why Wang turned to the model for this work.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u201cUltimately, we want to\u0026nbsp;make artificial intelligence and machine learning systems more transparent and understandable for non-technical users so, we wanted GAM Coach to be a glass box rather than a black box tool,\u201d said Wang.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u201cWe want people to be able to understand how and why a machine learning model makes a certain decision. We tailored our algorithm to integrate into a GAM because it is highly accurate, we know how it works, and we know how exactly how it makes predictions.\u201d\u003C\/p\u003E\r\n\r\n\u003Cp\u003EWang and his collaborators conducted an online user study of GAM Coach as part of the project. The team examined user logs from 41 Amazon Mechanical Turk workers to determine how everyday users would interact with the tool. The workers were presented with different loan scenarios and challenges, and then asked to use the tool to find recourse options that met their needs.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EAlong with a few minor usability issues, the researchers found that that personalized recourse plans are preferred over generic plans. They also found that users had a deeper understanding of how a decision was made and what they could do to change the outcome in the future.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EDespite the success of the tool so far, Wang says his team would need input from financial and legal experts before GAM Coach could be used in the real world. However, a demo and the code are available.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u201cDevelopers can also use our flexible Python library (`pip install gamcoach`) to generate recourse plans for GAMs,\u201d said Wang,\u0026nbsp;who is advised by School of CSE Associate Professor Polo Chau.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EHe is the lead author of\u0026nbsp;\u003Cem\u003EGAM Coach: Towards Interactive and User-centered Algorithmic Recourse\u003C\/em\u003E. The paper has been accepted and is being presented at the\u0026nbsp;\u003Ca href=\u0022https:\/\/chi2023.acm.org\/\u0022\u003E2023 ACM CHI Conference on Human Factors in Computing Systems\u003C\/a\u003E\u0026nbsp;later this month in Hamburg, Germany.\u003C\/p\u003E\r\n","summary":"","format":"limited_html"}],"field_subtitle":[{"value":"CSE Ph.D. student combines machine learning and data visualization to improve AI transparency"}],"field_summary":[{"value":"\u003Cp\u003EDeveloped at Georgia Tech, GAM Coach is the first interactive tool of its kind to give people with rejected loan applications the power to personalize recourse options that are realistically actionable to help ensure a better outcome in the future.\u003C\/p\u003E\r\n","format":"limited_html"}],"field_summary_sentence":[{"value":"New tool is the first of its kind to let everyday people create recourse strategies tailored to their unique needs. "}],"uid":"32045","created_gmt":"2023-04-14 14:34:58","changed_gmt":"2023-04-17 14:58:41","author":"Ben Snedeker","boilerplate_text":"","field_publication":"","field_article_url":"","dateline":{"date":"2023-04-17T00:00:00-04:00","iso_date":"2023-04-17T00:00:00-04:00","tz":"America\/New_York"},"extras":[],"hg_media":{"670547":{"id":"670547","type":"image","title":"loan_rejection image.jpg","body":null,"created":"1681482906","gmt_created":"2023-04-14 14:35:06","changed":"1681482906","gmt_changed":"2023-04-14 14:35:06","alt":"Stock image of young couple looking troubled with bills spread across the wooden floor","file":{"fid":"253428","name":"loan_rejection image.jpg","image_path":"\/sites\/default\/files\/2023\/04\/14\/loan_rejection%20image.jpg","image_full_path":"http:\/\/www.tlwarc.hg.gatech.edu\/\/sites\/default\/files\/2023\/04\/14\/loan_rejection%20image.jpg","mime":"image\/jpeg","size":144585,"path_740":"http:\/\/www.tlwarc.hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/2023\/04\/14\/loan_rejection%20image.jpg?itok=DM3waSkY"}},"670548":{"id":"670548","type":"image","title":"zijie-jay-wang-portrait.jpeg","body":"\u003Cp\u003EZijie (Jay) Wang is a Ph.D. student in Georgia Tech\u2019s School of Computational Science and Engineering (CSE).\u0026nbsp;\u003C\/p\u003E\r\n","created":"1681483036","gmt_created":"2023-04-14 14:37:16","changed":"1681483036","gmt_changed":"2023-04-14 14:37:16","alt":"Zijie (Jay) Wang, lead researcher and a Ph.D. student in Georgia Tech\u2019s School of Computational Science and Engineering (CSE).","file":{"fid":"253429","name":"zijie-jay-wang-portrait.jpeg","image_path":"\/sites\/default\/files\/2023\/04\/14\/zijie-jay-wang-portrait.jpeg","image_full_path":"http:\/\/www.tlwarc.hg.gatech.edu\/\/sites\/default\/files\/2023\/04\/14\/zijie-jay-wang-portrait.jpeg","mime":"image\/jpeg","size":161149,"path_740":"http:\/\/www.tlwarc.hg.gatech.edu\/sites\/default\/files\/styles\/740xx_scale\/public\/2023\/04\/14\/zijie-jay-wang-portrait.jpeg?itok=tghYnyEo"}}},"media_ids":["670547","670548"],"groups":[{"id":"47223","name":"College of Computing"},{"id":"50877","name":"School of Computational Science and Engineering"}],"categories":[{"id":"135","name":"Research"}],"keywords":[{"id":"2556","name":"artificial intelligence"},{"id":"9167","name":"machine learning"},{"id":"192525","name":"GAM Coach"},{"id":"192526","name":"algorithmic recourse"},{"id":"11559","name":"CSE computational science engineering"},{"id":"187915","name":"go-researchnews"}],"core_research_areas":[{"id":"39501","name":"People and Technology"}],"news_room_topics":[],"event_categories":[],"invited_audience":[],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[{"value":"\u003Cp\u003EBen Snedeker\u003Cbr \/\u003E\r\nCommunications Manager\u003Cbr \/\u003E\r\nCollege of Computing\u003Cbr \/\u003E\r\n\u003Ca href=\u0022albert.snedeker@cc.gatech.edu\u0022\u003Ealbert.snedeker@cc.gatech.edu\u003C\/a\u003E\u003C\/p\u003E\r\n","format":"limited_html"}],"email":[],"slides":[],"orientation":[],"userdata":""}}}