You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
<!--Please give us some feedback on your experience writing this PR ! https://app.datadoghq.com/forms/43db4c02-6837-400c-8083-692e141b1b88 !-->
### What does this PR do?
This uses anomaly scorer by default instead of time cluster which is now disabled by default.
### Motivation
We want to enable the scorer by default since we rely on it for smart adaptive sampling feature so our current direction is towards detection change points instead of sending events when we have bad behaviors.
### Describe how you validated your changes
Ran the benchmarks
### Additional Notes
Scenario | Before F1 (time cluster) | After F1 (scorer) | Δ F1
-- | -- | -- | --
block-building-outage | 0.285714 | 0.634606 | +0.348892
cascading-payment-failure | 0.063224 | 0.000083 | -0.063141
cassandra-repair-degradation | 0.227017 | 0.000000 | -0.227017
dns-upstream-outage | 0.054111 | 0.297905 | +0.243794
kafka-partition-saturation | 0.177494 | 0.577355 | +0.399861
lock-contention | 0.618715 | 0.032148 | -0.586567
memcached-saturation | 0.000006 | 0.557888 | +0.557882
pool-saturation | 0.000259 | 0.814789 | +0.814530
redis-cascade-billing | ~0 | ~0 | ~0
redis-cpu-saturation | 0.615742 | 0.016948 | -0.598794
tiered-cache-header-corruption | 0.001845 | ~0 | -0.001845
**Average** | 0.185830 | 0.266520 | +0.080690 (+43.42%)
Co-authored-by: celian.raimbault <celian.raimbault@datadoghq.com>
0 commit comments