In the example above, the suspected build cycle is with low confidence. Should we improve our algorithm for the determination of suspected build cycle as well?

It might worth a separate bug to track that improvement.

For this particular example, if we can confirm that build 34108 is really stable, we should get a different suspected build with different confidence. Although I agree we need to continue improving our algorithms to improve the correctness of suspected build

OK, then let's revisit after build 34108 is verified.

The following revision refers to this bug:
  https://chromium.googlesource.com/infra/infra/+/aea7c3a7665ac7c0b52ced25a0d70785ccb56b88

commit aea7c3a7665ac7c0b52ced25a0d70785ccb56b88
Author: Chan <chanli@chromium.org>
Date: Thu Mar 09 23:18:53 2017

[Findit] Flake Analyzer:Rerun swarming tasks on stable build instead of going back directly.

If swarming rerun result is stable on a build, rerun the task with higher iterations
to either confirm this build is really stable to start sequential search or get flaky
result on a bigger iteration number.

After this change:
1. no need to check stable region
2. no need to check if test started to be flaky or maintain flaky region: every flaky  result can be treated equally

BUG= 695528 

Change-Id: Idbd5ee4d40967589f8bf1177e89dd9bb4db52aab
Reviewed-on: https://chromium-review.googlesource.com/447198
Commit-Queue: Chan Li <chanli@chromium.org>
Reviewed-by: Jeffrey Li <lijeffrey@chromium.org>

[modify] https://crrev.com/aea7c3a7665ac7c0b52ced25a0d70785ccb56b88/appengine/findit/waterfall/flake/test/recursive_flake_pipeline_test.py
[modify] https://crrev.com/aea7c3a7665ac7c0b52ced25a0d70785ccb56b88/appengine/findit/waterfall/flake/recursive_flake_pipeline.py
[modify] https://crrev.com/aea7c3a7665ac7c0b52ced25a0d70785ccb56b88/appengine/findit/waterfall/flake/lookback_algorithm.py
[modify] https://crrev.com/aea7c3a7665ac7c0b52ced25a0d70785ccb56b88/appengine/findit/waterfall/flake/test/lookback_algorithm_test.py