September 26-27, 2016
New York, NY

Combining statistics and expert human judgement for better recommendations

Jianqiang (Jay) Wang (Stitch Fix), Jasmine Nettiksimmons (Stitch Fix)
4:35pm–5:15pm Tuesday, 09/27/2016
Interacting with AI
Location: 3D08 Level: Intermediate

Prerequisite knowledge

  • A basic knowledge of probability and statistics and the retail industry
  • What you'll learn

  • Explore a real-world use case combining AI and human experts for a personalized shopping experience
  • Learn the interesting aspects of designing a recommendation system with implicit feedback and the challenges in interacting with an AI system to increase credibility and collaboration
  • Description

    Jay Wang and Jasmine Nettiksimmons explore the business model of Stitch Fix, an emerging startup that uses artificial intelligence and human experts for a personalized shopping experience. Stitch Fix’s service combines recommendation algorithm and human stylists in curating clothes for customers. Jay and Jasmine discuss data collection and feature engineering for the recommendation algorithm, as well as some algorithmic innovations. They then highlight the challenges encountered implementing Stitch Fix’s recommendation algorithm and interacting AI with human stylists before briefly introducing other problems Stitch Fix’s data science team is solving, including language processing, computer vision, inventory simulation, and demand forecasting.

    Photo of Jianqiang (Jay) Wang

    Jianqiang (Jay) Wang

    Stitch Fix

    Jianqiang “Jay” Wang is a data science lead at Stitch Fix working on recommendation algorithms and human computer interaction. Previously, Jay worked in academia on survey sampling, nonparametric smoothing, and Bayesian hierarchical models; at HP Labs on demand forecasting and supply-chain management; and as a data scientist at Twitter on ads CTR prediction and ranking. Jay holds a PhD in statistics from Iowa State University.

    Photo of Jasmine Nettiksimmons

    Jasmine Nettiksimmons

    Stitch Fix

    Jasmine Nettiksimmons is a data scientist at Stitch Fix, where she focuses on robust parameter estimation in observational data and assessing how successfully humans interact with a live recommendation system. Prior to joining Stitch Fix, she worked in the field of cognitive aging with research focusing on biomarker profiles which are predictive of cognitive decline and dementia. In addition to her work in cognitive aging, she has a broad publication record across many public health and social issues including rural health care delivery, childhood obesity, domestic violence prevention, and family-friendly policy usage. Jasmine holds a PhD in epidemiology from UC Davis.