Machine learning is all the rage, but very few shops understand enough about it to implement an effective solution. Machine learning involves having an expert on the team that understands all the mathematical algorithms used to analyze data and map it back to users. Explaining machine learning to the masses, has often reduced complex programs to mere ‘Related Articles’ or ‘Books you might like’ examples. Developers new to machine learning, then gain a false sense of confidence after home-brewing a recommender engine – but they’ve only scratched the surface.
Prediction IO takes all the complex math out of the equation and bundles a machine learning server into an open source platform to store user data. Here are some screenshots that Shekhar Gulati posted to illustrate the basics of Prediction IO.
4. Collect the Data and Train the Model
The adding the engine will requires setting up the Prediction IO server, then writing client code to store user data and attributes to the server, then retrieving recommended data. This solution puts machine learning into the hands of Web developers without the need for a math programmer guru. It seems like a good start, but will still require many hands on deck to get something like this rolled out with any longterm meaning.