Learning Google Prediction API with JSON data

Question

I've found some samples of using Google Prediction API, but there are only CSV data used in sample. I want to feed Prediction API with data organized like this:

[{
    number: 1,
    type: 0,
    channels: [
        [1, 5, 3, ...], 
        [7, 1, 3, ...], 
        [1, 23, 7, ...], 
        [5, 4, 3, ...], 
        [1, 3, 3, ...], 
        [6, 4, 3, ...]]
}, ...]

Is that possible to learn the model with such data or I have to use CSV only?

Answer 1

No. The prediction API currently has support for only csv format. You could easily write a script that iterates through all instances in the JSON array and converts them into comma-separated values.