Moving window online stats about onlinestats.jl HOT 3 CLOSED

Thanks for the interest. This is a duplicate of #20, so I'm going to close it. As I said there, moving windows aren't exactly "online statistics" in the sense that you need a history of data to compute, but I do agree they are important to support. We have the pieces, just need to put them together nicely.

from onlinestats.jl.

Thanks for the reply but I am not sure if that is true.

You can compute online moving window in 2 ways.

One is to wait for the circular buffer to fill and then start computing the stats.

The second is to start right away based on the window size and increase it to its maximum size.

For example, the when 1 sample appears the window size is 1. When two samples appear the window size is 2.
For example [1] appears mean is 1
then [2] appears and circular buffer has [1 2] and mean is 1.5

The two approaches converge to the same value once the circlular buffer is full.

Thanks for the package by the way. Even though I dont use it now I really appreciate it and might use it in the future.

from onlinestats.jl.

Fair enough, but that's a small implementation detail. We could support both.

On Aug 28, 2015, at 1:24 AM, NowItIsTime [email protected] wrote:

Thanks for the reply but I am not sure if that is true.

You can compute online moving window in 2 ways.

One is to wait for the circular buffer to fill and then start computing the stats.

The second is to start right away based on the window size and increase it to its maximum size.

For example, the when 1 sample appears the window size is 1. When two samples appear the window size is 2.
For example [1] appears mean is 1
then [2] appears and circular buffer has [1 2] and mean is 1.5

The two approaches converge to the same value once the circlular buffer is full.

Thanks for the package by the way. Even though I dont use it now I really appreciate it and might use it in the future.

—
Reply to this email directly or view it on GitHub.

from onlinestats.jl.