How is this API implemented in AML?

Nov 21, 2014 at 6:40 AM
If I understand correctly, AML can only publish stateless web service APIs.

But for time series, usually you may need to check consecutive data points and find the correlation between them.

I noticed the source code for input data looks like this.

var series = "9/21/2014 11:05:00 AM=3;9/21/2014 11:10:00 AM=9.09;9/21/2014 11:15:00 AM=0;";

Does this mean that I need to send more than one data point to this API for every prediction? If that's the case,
  1. How is the number of data points determined?
  2. Looks like there will be lots of overlapping data between two consecutive predictions and also the two consecutive predictions are actually "independent" from each other?
For Example. data series might looks like this:

0 s: 3.1
1 s: 4.2
2 s: 4.7
3 s: 4.9
4 s: 2.0
5 s: 1.9
6 s: 2.2
...

If I decide to predict using 5 consecutive points, I need to make the following calls to this API:

3.1 4.2 4.7 4.9 2.0
4.2 4.7 4.9 2.0 1.9
4.7 4.9 2.0 1.9 2.2
...

Am I understanding this API correctly? Thanks a lot!