tangdian commented on a change in pull request #4149: [TE] support monitoring
granularity for Holt Winters
URL: https://github.com/apache/incubator-pinot/pull/4149#discussion_r277110926
##########
File path:
thirdeye/thirdeye-pinot/src/main/java/org/apache/pinot/thirdeye/detection/components/HoltWintersDetector.java
##########
@@ -112,13 +112,25 @@ public void init(HoltWintersDetectorSpec spec,
InputDataFetcher dataFetcher) {
this.pattern = spec.getPattern();
this.smoothing = spec.getSmoothing();
this.sensitivity = spec.getSensitivity();
+ this.monitoringGranularity = spec.getMonitoringGranularity();
+
+ if (this.monitoringGranularity.equals("1_MONTHS")) {
+ this.timeGranularity = MetricSlice.NATIVE_GRANULARITY;
Review comment:
> Pinot doesn't support monthly aggregation from the query. Monthly
detection is only doable if only the dataset itself is marked as daily
granularity but only have one data point per month. So for monthly detection,
when fetching data should still use native granularity, but the algorithm needs
to be aware that it's a monthly granularity data and fetches the right training
data to construct the correct baseline.
>
> @tangdian I'm not sure this will work for the monthly dataset. looks like
you are still retrieving 60 days' training data. Then it's likely that there
only 1 data point in there. Since this is not required, you can throw a
NotImpelemented exception in this case if it's hard to make it work.
I have updated the code to support 1_MONTHS data
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
