If I have a graph of customers and what products they've bought, Directed Edge makes it easy to figure out what products to recommend to each customer. Generating the same recommendations with Google's prediction API is not as obvious to me. What data do you train it with? The nth product the customer bought as the output, and the 1..n-1th products as the input?
Maybe I'm slow, but I don't see how google's prediction API is a good replacement for collaborative filtering type recommendation engines.
Comments
If I have a graph of customers and what products they've bought, Directed Edge makes it easy to figure out what products to recommend to each customer. Generating the same recommendations with Google's prediction API is not as obvious to me. What data do you train it with? The nth product the customer bought as the output, and the 1..n-1th products as the input?
Maybe I'm slow, but I don't see how google's prediction API is a good replacement for collaborative filtering type recommendation engines.