Predictive Analytics Times
Predictive Analytics Times
EXCLUSIVE HIGHLIGHTS
Exploring the Increased Significance of Third Party Data Within Big Data
 For today’s leading deep learning methods...
Dr. Data Show Video: Five Reasons Computers Predict When You’ll Die
  Watch the latest episode of...
How to Perfect Your Data Science Resume in 3 Easy Steps
 Breaking into the world of Data...
Robotic Modeler for Marketing Campaigns
 SUMMARY: Automated modeling is already in...
SHARE THIS:

3 years ago
Wise Practitioner – Predictive Analytics Interview Series: Mario Vinasco at Facebook

 

In anticipation of his upcoming conference presentation, Advanced Mario V imageExperimentation in Social Networks at Predictive Analytics World San Francisco, April 3-7, 2016, we asked Mario Vinasco, Marketing Analytics Data Scientist at Facebook, a few questions about his work in predictive analytics.

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

A:  We use predictive models for several different purposes mainly in the context of advertising campaigns and sentiment measurement.

  • To identify lookalikes: these models take demographics and behaviors and identify users with similar characteristics.
  • To partition the social graph into clusters of connected users to run network experiments
  • To attribute product usage to a campaign: there are many sources of bias in experiments, for example the Ad optimizer bias the treated or exposed group while the control group remains intact; we use inverse propensity and regression models to rebalance.
  • To rebalance survey responses between tests and control groups due to biases induced by the Ad delivery engine.

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

A: We can accurately attribute and quantify campaign investment to outcomes and can create target audiences where we can increase either product usage or sentiment.

We have invested in markets and hero products accordingly.

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

A: In a recent network based experiment, we were able to attribute the ROI of a very large ad campaign whose message went viral, and because of these network effects a traditional A|B test was not possible.

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

A: It has been very interesting to see that the law of long tails apply over and over; it can be product usage, likes received on a pots, etc., and this illustrates how deceiving simple averages can be.

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

A:  People will understand better some of the dynamics of social networks, especially experimentation set up.

———————

Don’t miss Mario’s conference presentation, Advanced Experimentation in Social Networks, on Monday, April 4, 2016 at 4:45 to 5:30 pm at Predictive Analytics World San Francisco. Click here to register to attend. USE CODE PATIMES16 for 15% off current prices (excludes workshops).

By: Eric Siegel, Founder, Predictive Analytics World

Eric Siegel is the founder of Predictive Analytics World (www.pawcon.com) — the leading cross-vendor conference series consisting of 10 annual events in New York, Chicago, San Francisco, Washington DC, London, and Berlin — and the author of the award-winning book, Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die – Revised and Updated Edition, (Wiley, 2016).

Leave a Reply