vividcortex / ewma Goto Github PK

For some purposes, it's important to be able to set the EWMA's value directly. For example, suppose you keep track of a EWMA and then save it to a database, and later retrieve it and continue to add more values. (We do this at VividCortex.)

We should enhance the interface with an Add(float64) function, and implement these functions in our two public implementations. The semantics of Add should be to set the value and mark the EWMA as initialized, in case it has internal tracking of whether it's "warmed up" or not.

These properties let it use less memory. It will also behave differently when it's equal to zero, which is assumed to mean uninitialized, so if a value is likely to actually become zero over time, then any non-zero value will cause a sharp jump instead of a small change.

The disadvantages for SimpleEWMA like the sensitivity for 0 values could be solved by using *float64 for storing the value internally. Could do a PR if desired.

github.com/VividCortex/ewma 1.1.1 does not pass unit tests with Go 1.10:

+ GOPATH=/builddir/build/BUILD/ewma-1.1.1/_build:/usr/share/gocode
+ go test -buildmode pie -compiler gc -ldflags '-extldflags '\''-Wl,-z,relro  '\'''
# github.com/VividCortex/ewma
./ewma_test.go:39: Errorf format %d has arg e.Value() of wrong type float64
./ewma_test.go:53: Errorf format %d has arg e.Value() of wrong type float64
./ewma_test.go:83: Errorf format %d has arg e.Value() of wrong type float64
FAIL    github.com/VividCortex/ewma [build failed]

Useful for dependency managers like dep.

I was really looking forward to seeing the picture of the moving average cat.

The readme says

So, for example, if you have a time-series with samples once per second, and you want to get the moving average over the previous minute, you should use an alpha of .032786885. This, by the way, is the constant alpha used for this repository's SimpleEWMA.

That alpha corresponds to N = 60, 2/(N+1) = 2/61 = .032786885.

The actual code uses 2/31:

The readme may be using N with two different meanings in the section on "Choosing Alpha", but I don't think so. I think 2/31 is incorrect.

The comment in the code is also inconsistent with the readme. The readme talks about "N samples", whereas the code talks about "average age".

It's certainly possible I'm misunderstanding something.

Before you file an issue, please consider:

We only accept pull requests for minor fixes or improvements. This includes:

• Small bug fixes
• Typos
• Documentation or comments

Please open issues to discuss new features. Pull requests for new features will be rejected,
so we recommend forking the repository and making changes in your fork for your use case.

I'm assuming the current state is stable. Could you please add a tag/release? This would make it easier for specifying versions for dep.

Hello,

I am in the process of packaging ewma for Fedora but I encountered problems running the tests. On both ppc64le and s390x arches, the following test fails:

+ export GOPATH=/builddir/build/BUILDROOT/golang-github-VividCortex-ewma-0-0.1.git4cc8cc5.fc27.ppc64le//usr/share/gocode:/usr/share/gocode
+ GOPATH=/builddir/build/BUILDROOT/golang-github-VividCortex-ewma-0-0.1.git4cc8cc5.fc27.ppc64le//usr/share/gocode:/usr/share/gocode
+ go test -compiler gc -ldflags '' github.com/VividCortex/ewma
--- FAIL: TestSimpleEWMA (0.00s)
	ewma_test.go:26: e.Value() is 4734.500946466117, wanted 4734.500946466118
--- FAIL: TestVariableEWMA (0.00s)
	ewma_test.go:40: e.Value() is 4734.500946466117, wanted 4734.500946466118
FAIL
FAIL	github.com/VividCortex/ewma	0.018s

Could you identify why there is a discrepancy in the value given versus the value expected, only on these two arches while others arches test are fine?