WIP: Metric anomaly check #103
Conversation
| MetricsCalculator.calculateMetrics(describedDataset, metricDescriptors) | ||
| (describedDataset, metricValues) | ||
| } | ||
| val allPreviousMetricsFut: Future[Map[Instant, Map[SingleDsDescription, Map[SimpleMetricDescriptor, MetricValue]]]] = |
There was a problem hiding this comment.
We should load only historic metrics that we actually need
|
|
||
| for { | ||
| _ <- storedMetricsFut | ||
| allPreviousMetrics: Map[Instant, Map[SingleDsDescription, Map[SimpleMetricDescriptor, MetricValue]]] <- allPreviousMetricsFut |
There was a problem hiding this comment.
As per above we should only load historic metrics we actually need
timgent
force-pushed
the
metric-anomaly-check
branch
from
69802f8 to
ba46206
Compare
|
That would be a great feature. What's left for this to get merged? |
|
@kingbrown9 in the current form it could be merged with not too much more work - would just need to fully review the code. The reason it stalled was because there's an issue which is trickier to solve. When you get an anomalous result should future anomaly checks take that into account or not? There's no way to know without human intervention to say if it was legitimate or not. This was part of the reason for starting to work on a UI (which is still in a very early, basic state, though did work in a simple form with the API, though it didn't cover this use case yet). Thoughts appreciated though if it's useful even in it's current form as probably not too much work to get this merged. |
No description provided.