AI in Financial Services–Final Mile of a Debit Card Fraud Machine Learning Model
This meetup was recorded on August 21, 2018 in Mountain View, CA.
Description:
Today, credit and debit card fraud detection machine learning models are a critical component of a financial institution’s fraud mitigation operations. Predictive performance of these models is extremely important to help catch fraudsters and shut down a customer’s compromised card as soon as possible. Because of this, data scientists often focus all efforts on the training phase of the model life cycle, trying to squeeze out as much predictive power as possible. In highly regulated U.S. banks, and really anywhere one is deploying machine learning models for critical business results, carefully delivering the models that final mile into production can be just as important. In this talk, we explore two ways data scientists can help deliver in the final mile: Gradient Boosting Machine (GBM) fraud model interpretability and model monitoring.
Speaker's Bio:
Daniel Dixon is a senior data engineer on the Enterprise Analytics & Data Science team at Wells Fargo, where he is responsible for designing and building scalable, big data pipelines to feed intelligent systems across the bank. In this role he specializes in big data and advanced analytics challenges, utilizing machine learning, statistics, process optimization, and visualization techniques to analyze and assemble large, complex datasets. Prior to joining Wells Fargo in 2014, Daniel spent five years as a professional services consultant for Teradata with a focus on visualization and ETL technologies. He holds a Bachelor of Science in Electrical Engineering with a minor in Computer Science from the Georgia Institute of Technology.
LinkedIn: www.linkedin.com/in/danielbdixon
This meetup was recorded on August 21, 2018 in Mountain View, CA.
Description:
Today, credit and debit card fraud detection machine learning models are a critical component of a financial institution’s fraud mitigation operations. Predictive performance of these models is extremely important to help catch fraudsters and shut down a customer’s compromised card as soon as possible. Because of this, data scientists often focus all efforts on the training phase of the model life cycle, trying to squeeze out as much predictive power as possible. In highly regulated U.S. banks, and really anywhere one is deploying machine learning models for critical business results, carefully delivering the models that final mile into production can be just as important. In this talk, we explore two ways data scientists can help deliver in the final mile: Gradient Boosting Machine (GBM) fraud model interpretability and model monitoring.
Speaker’s Bio:
Daniel Dixon is a senior data engineer on the Enterprise Analytics & Data Science team at Wells Fargo, where he is responsible for designing and building scalable, big data pipelines to feed intelligent systems across the bank. In this role he specializes in big data and advanced analytics challenges, utilizing machine learning, statistics, process optimization, and visualization techniques to analyze and assemble large, complex datasets. Prior to joining Wells Fargo in 2014, Daniel spent five years as a professional services consultant for Teradata with a focus on visualization and ETL technologies. He holds a Bachelor of Science in Electrical Engineering with a minor in Computer Science from the Georgia Institute of Technology.
LinkedIn: www.linkedin.com/in/danielbdixon
