Machine Learning Times
Machine Learning Times
EXCLUSIVE HIGHLIGHTS
Effective Machine Learning Needs Leadership — Not AI Hype
 Originally published in BigThink, Feb 12, 2024.  Excerpted from The...
Today’s AI Won’t Radically Transform Society, But It’s Already Reshaping Business
 Originally published in Fast Company, Jan 5, 2024. Eric...
Calculating Customer Potential with Share of Wallet
 No question about it: We, as consumers have our...
A University Curriculum Supplement to Teach a Business Framework for ML Deployment
    In 2023, as a visiting analytics professor...
SHARE THIS:

5 years ago
Wise Practitioner – Predictive Analytics Interview Series: Zeydy Ortiz at DataCrunch Lab

 

By: Eric Siegel, Program Co-Chair, Predictive Analytics World for Financial

In anticipation of her upcoming conference presentation at Predictive Analytics World for Financial Las Vegas, June 16-20, 2019, we asked Zeydy Ortiz, CEO at DataCrunch Lab, a few questions about their deployment of predictive analytics. Catch a glimpse of her presentation, Will They Stay Or Will They Go? A Customer Lifetime Value Case Study, and see what’s in store at the PAW Financial conference in Las Vegas.

Q: In your work with predictive analytics, what behavior or outcome do your models predict?

A: We work with businesses in different industries (IT, finance, retail, manufacturing, etc.) to predict outcomes that are relevant to the business. I started my career developing predictive models of server performance and energy efficiency to inform business strategy at IBM. When working on new projects, the particular behavior or outcome to predict is determined by a combination of the business objectives, the data available, and how that model would be used.

At the Predictive Analytics World conference, I will share a case study on predicting how long a customer would stay in a program that can be used by an organization to estimate the customer life time value for each of their new customers. Due to differences in the organizations adopting the predictive models, we had to use different deployment methods which in turn affected model selection.

Q: How does predictive analytics deliver value at your organization – what is one specific way in which it actively drives decisions or operations?

A: Predictive analytics enables companies to decide on the best way to serve their customers based on their expected customer lifetime value (CLTV). Before building predictive models, the organizations that we worked with had done extensive mining of their historical records. They knew how many customers they were serving at any given time, the demographic characteristics of the customers they were attracting, and how long they stayed in the program on average. With these descriptive analytics they can determine the average CLTV for the population of customers. However, they can not personalize the service for a new customer. Using a predictive model they are now able to build different programs that would help them better attract and retain customers.

Q: Can you describe a quantitative result, such as the predictive lift of your model or the ROI of an analytics initiative?

A: A customer that is retained for an additional 2 months would bring in 12% additional revenue to the organization. The predictive model informs counselors how to best serve new customers based on their expected CLTV.

Q: What surprising discovery or insight have you unearthed in your data?

A: When we analyzed data and developed models for three different organizations, we found evidence of the “No Free Lunch” (NFL) theorem. The essence of the NFL theorem is that there is no machine learning model that works best for every problem. We experimented building models for these organizations using a variety of machine learning algorithms: survival analysis, poisson linear regression, bayesian linear regression, decision forest regression, boosted decision tree regression, gradient boosting machine regression, and neural network regression. While we were able to use a survival model to predict how long a customer would stay in the program for two of the organizations, a boosted decision tree regression algorithm worked significantly better for the data from the third organization. The only way for us to know what worked best was to build experiments based on sound methodology and test it out.

Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.

A: One of our key tenets is to “start with the end in mind.” Developing a useful end-to-end predictive analytics model requires asking detailed questions before, during, and after modeling. The end result depends on the business objectives, the data that is available, and how the model will be used.

—————————–

Don’t miss Zeydy’s presentation, Will They Stay Or Will They Go? A Customer Lifetime Value Case Study, at PAW Financial on Wednesday, June 19, 2019 from 3:55 to 4:15 PM. Click here to register for attendance.

By: Eric Siegel, Program Co-Chair, Predictive Analytics World for Financial

This content is restricted to site members. If you are an existing user, please log in on the right (desktop) or below (mobile). If not, register today and gain free access to original content and industry news. See the details here.

Comments are closed.