How Game Theory and the Shapley Value Apply to Predictive SEO

Learn How Predictive SEO Impacts the Future of SEO with Rebecca Berbel 07 December 2021 at 17 H CEST (Paris)

In this clip, Rebecca Berbel explains how Game Theory and in particular the Shapley value can be usefully applied to Machine Learning models for predictive SEO. Listen to her explain, and you’ll understand!

00:00 How are you going to rank?
00:13 Machine Learning model test predictions for all features
00:25 Needing complete set of impactful features
00:53 Shapley value won a Nobel Prize

Transcript from: How Game Theory and the Shapley Value Apply to Predictive SEO

No, no, no. You can’t have your citation flow. How are you going to rank now? Now you can’t have your other thing. How are you going to rank now?

And (the ML model) is going to test its predictions through all of that and compute that for all of the possible subsets of features. So you need a complete set of features or at least a complete set of significantly impactful features. So it depends on the features you choose to analyze because we don’t know all ranking signals.

Yeah. And just to cut in there, I’m really shocked. You’ve said it depends at least five times when Anton hasn’t put up his sign.

I’m very disappointed.

But my point here is that Shapley won a Nobel Prize for this type of thinking in 2012. Some of the Python libraries for this came out in 2017.

Similar Posts