How much data does PredictionIO need to start predicting?

Question

data mining

Hello.
Would like to know if anyone has any experience with PredictionIO .
More precisely, I wanted to know:
- how much data should be fed to it so that it really starts to predict something?
- what kind of hardware is needed for work (turnover of about 100k users per day)?
- where can I read (or who can I ask) about the experience of the real implementation of this contraption?
Thanks in advance.

Answer 1

I'll answer myself :)
- There may not be much data - the main thing is to have more U2I actions. Ie this is the number of user actions to the items that are loaded there. The larger this ratio, the better the predictions will be.
- Iron on the first is the most common, single node hadup instance - of course, it eats the CPU, but the usual Dedik (Core i7, 32GB) is still enough.
- The documentation is normal on the site, there is a Google group for questions https://groups.google.com/forum/?hl=en#!forum/pred...