Webinar June 18: Deep Learning Approaches to Forecasting and Planning Register

Demystifying Data Science Talk Recap: Atif Kureishy on Applying AI to Understand Customer Behaviors

By Emily Wilson • April 23, 2020

This is the 7th installment in an ongoing blog series to recap the talks given on Day 2 of our 2019 Demystifying Data Science live online conference. The content presented on Day 2 was designed for Business Leaders, Managers, and Practitioners. 

During his talk, Atif Kureishy (Global Vice President of AI & Deep Learning Products at Teradata) discusses how to merge offline and online activity in order to better serve customers while staying cost-efficient in the retail space. 

While there has been significant investment in artificial intelligence across industries, there's still room for growth from the "observational" perspective, according to Kureishy. To explain what that means, he starts by showcasing how he and his team at Teradata think about AI. 

This approach allows the company to "drive better answers" that benefit retail operations across the board. How so? It all starts with data, which Kureishy describes as the "underpinning of AI." He breaks down data in the following three ways: 

1) Transactional: data that’s highly curated and has been driving value in the corporate environment for around 40 years. 

2) Interactional: human-generated data derived from mobile and social media, which has developed in the last 20 years. 

3) Observational: over the coming decades, companies will be focused on this data, which comes from imagery, video, and more and allows for the observation of human behavior.

Where can you see examples of observational data today?

"It's everywhere," according to Kureishy. You can find it in security, manufacturing, drones, semi-autonomous cars, and so forth. 

"We now have all of this rich observational data available to us in the enterprise and it's waiting to be analyzed," he adds. 

In the following, you see a screenshot of a video that Kureishy shows to demonstrate how observational data is being collected and analyzed in the retail sector. Customers are categorized as they interact with the products and store layout, and at the bottom of the image, moving from "walking" to "engaging" in real-time.

As Kureishy explains in detail, it's important to note that faces are blurred and no biometrics are used in these processes. 

"We have to be fixated on the privacy, ethics, and compliance piece," he said. "Controls put in place and thought and privacy and the ethical aspects of all of this need to be first-class considerations."

These privacy-related concerns are the base of Teradata's observational data work

In the end, this data results in time-series data, according to Kureishy.

"For each time stamp on every camera, we see a subject, and we can locate them in an X and Y coordinate," he said. In effect, we can see what it looks like for Person A to move from "walking" to "engagement" and have a time-series record of it. 

And while that's "interesting," according to Kureishy, it's not helpful unless it can be put into context.

This is where the offline and online worlds meet, and where you can start answering questions like "How did employee or associate engagement influence conversion in the store?" or "How did a specific promotion change customer behavior in the store?"

"All of that unlocks given that we can now process this observational data," says Kureishy. And this can start influencing decision making for the store. 

So how can retailers take steps to improve customer satisfaction based on this data? How can they improve the cost to serve?

To get robust answers, watch his full presentation, including a lively Q&A discussion. Register for free here and receive a link via email for access to this and all other Demystifying Data Science talks.


Similar Posts

business resource
Test Your Data Literacy Daily with Our New #DemystifyDS Question of the Day

By Metis • April 28, 2020

Our mission at Metis is to help people find value in data, and we're excited to announce our new data literacy initiative, the Demystifying Data Science Question of the Day. With each question answered and explained, you'll begin strengthening your data literacy muscles.

business resource
Updates to the Metis Corporate Training Page Break Down Services & Courses

By Metis • May 14, 2020

On our new Corporate Training page, we provide an overview of our offerings, complete with a breakdown of all services and courses, which include Data Literacy, Python for Data Analysts, Machine Learning Foundations, and more. We also wanted to highlight our history, backed by Kaplan, and our team, made up of smart, talented, and ambitious educators and thinkers.

business resource
Understanding Natural Language Processing

By Tony Yiu • May 22, 2020

Natural language processing (NLP) is one of the trendier areas of data science. Its end applications are many – chatbots, recommender systems, search, virtual assistants, etc. In this post, Data Scientist Tony Yiu helps readers understand the basics of NLP, including why it's important and how it helps us.