R is very powerful, but its use is mostly limited to statisticians. So providing a simple interface for people to leverage predictive models built in R is a great idea.
But their sample output (http://predict.i2pi.com/8MjSV2AA) just shows the distributions of all the input variables. I don't actually see any predictions.
Edit: There are no predictions there because there are no "NA"'s in the sample file to predict. You would think they'd create a demo file that would showcase the predictions!
Could people making things like this make it more accessible by, say, hooking it up to Google Docs? Or some other web spreadsheet? "Point us at your data and watch us do cool stuff with it."
Forcing users to create a specially formatted CSV from scratch just to try something out is asking for low participation.
Good point. My concern here is that when people work with spreadsheets they tend to violate they layout principles that I'd prefer data to adhere to. Maybe a better option would be to create a google docs plugin that lets people clean up their spreadsheets - interactively pointing out where cleaning needs to be done.
Hi - The interface is not very intuitive. I can say that with some authority as I'm the guy behind this. The idea was to get it out and see how people use it.
And so far I'm seeing that many people are getting confused by the correct file format and how to annotate predictions. But given that so many people are giving it a go, I'll spend some time cleaning it up and making it more user friendly.
But their sample output (http://predict.i2pi.com/8MjSV2AA) just shows the distributions of all the input variables. I don't actually see any predictions.
Edit: There are no predictions there because there are no "NA"'s in the sample file to predict. You would think they'd create a demo file that would showcase the predictions!