Guess.js: Libraries for enabling Machine Learning driven experience on the web

github.com

61 points·pestkranker··8 comments

Would be great to have an example of Guess.js with https://github.com/sveltejs/sapper

Cool idea and a great application of predictive machine learning. Wise to tie into Google analytics. Wonder how long until Google acquires them.

Thanks for the thumbs up! We have an abstract report provider which allows developers to use their favorite analytics platform easily.

Google Analytics is quite convenient - it allows the reuse of already existing data, and does not require the management of an analytics service.

I'd love to see some examples of this in action. I'm not the best at implementing these kinds of things without seeeing examples.

We have a few examples in the readme. You can find demo applications using the Guess.js plugin for webpack here https://github.com/guess-js/guess/tree/master/packages/guess...

The most real-world demo is the Gatsby Wikipedia example which Addy Osmani demonstrated on Google I/O. You can find it here at https://guess-gatsby-wikipedia-demo.firebaseapp.com/. You can toggle link highlighting by pressing "h". Red links are very likely to be visited, orange/yellow ones have mild probability, and green ones are unlikely. The highlighting is based on the model that Guess.js builds from the Google Analytics report.

Here's demo of using Guess.js with static sites https://github.com/guess-js/guess/tree/master/experiments/gu...

I think there are some issues with the demo, there are 502 errors (CORS) coming up in the console.

Is this a desktop-only demo?

https://github.com/mgechev/guess-next for predictive prefetching of JavaScript bundles

Reply on news.ycombinator.com