US 12,321,567 B2
Performing an action based on user interaction data
Rohit Sarin, McLean, VA (US); Ryan Tomlinson, McLean, VA (US); Elizabeth Maue, McLean, VA (US); Alex Baird, McLean, VA (US); Jose Martinez, McLean, VA (US); Chengjun Zhu, McLean, VA (US); Weiyan Sun, McLean, VA (US); and Robert Cezeaux, McLean, VA (US)
Assigned to Capital One Services, LLC, McLean, VA (US)
Filed by Capital One Services, LLC, McLean, VA (US)
Filed on Dec. 28, 2020, as Appl. No. 17/247,852.
Application 17/247,852 is a continuation of application No. 16/284,616, filed on Feb. 25, 2019, granted, now 10,877,632.
Prior Publication US 2021/0117055 A1, Apr. 22, 2021
Int. Cl. G06F 18/2113 (2023.01); G06F 3/0482 (2013.01); G06F 3/04883 (2022.01); G06F 3/16 (2006.01); G06N 3/08 (2023.01); G06N 20/00 (2019.01)
CPC G06F 3/0482 (2013.01) [G06F 3/04883 (2013.01); G06F 3/167 (2013.01); G06F 18/2113 (2023.01); G06N 3/08 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A method, comprising:
obtaining, by a device, historical interaction data concerning a plurality of user interactions;
obtaining, by the device, historical response data concerning a plurality of user responses associated with the plurality of user interactions;
determining, by the device, a respective user interaction score for each user interaction of the plurality of user interactions;
processing, by the device, the historical interaction data and the historical response data, using a pattern mining technique, to determine one or more touchpoint sets associated with the plurality of user interactions;
processing, by the device, the historical interaction data and the historical response data, using at least one of a recursive feature elimination technique or a gradient boosting technique, to generate a machine learning model,
the machine learning model being trained to determine associations between the respective user interaction scores and the one or more touchpoint sets, and
the machine learning model being trained to determine to what degree the one or more touchpoint sets affect the respective user interaction scores;
processing, by the device, information concerning a current user interaction using the machine learning model to determine a particular touchpoint set of a plurality of touchpoint sets associated with the current user interaction;
ranking, by the device, the plurality of touchpoint sets associated with the current user interaction based on respective estimated user interaction scores associated with each touchpoint set, of the plurality of touchpoint sets associated with the current user interaction;
identifying, by the device, using the machine learning model, and based on the ranking, the particular touchpoint set, when the particular touchpoint set is associated with a respective estimated user interaction score, of the respective estimated user interaction scores, that is greater than a user interaction score associated with the current user interaction; and
performing, by the device, one or more actions based on the particular touchpoint set.