LowPtElectrons: ID models for 10_6_X

[bainbrid]

This PR contains models that are required by PR cms-sw/cmssw#32372 (which is a back port of cms-sw/cmssw#31220).

There are two weights files:

RunII_Autumn18_LowPtElectrons_mva_id.xml.gz

• this is the model used by RECO throughout the 10_6_X release cycle

LowPtElectrons_ID_2020Sept15.root

• this is the model to be used for reminiAOD UL (via the run2_miniAOD_UL modifier)
• is the same model used in Add low-pT electrons to MINIAOD, update ID, improve end user experience cms-sw/cmssw#31220
• has depth=15, trees=1000, extra variables, no regression, base on BParking samples.

[cmsbuild]

A new Pull Request was created by @bainbrid for branch master.

@perrotta, @smuzaffar, @mrodozov, @cmsbuild, @slava77, @jpata can you please review it and eventually sign? Thanks.
@Sam-Harper, @jainshilpi, @lgray, @sobhatta, @afiqaize, @varuns23, @ram1123 this is something you requested to watch as well.
cms-bot commands are listed here

[slava77]

I'm confused, what is the difference between this PR and #15?

Both are a single commit and both add a file LowPtElectrons/LowPtElectrons_ID_2020Sept15.root
... and #15 is already merged

[bainbrid]

In response to: #17 (comment):

In short, this commit changed the RunII_Autumn18_LowPtElectrons_mva_id.xml.gz file. Here's the history for the XML file. This PR reverts that change.

It's difficult to know how to navigate GitHub to find the commit or tag used to build a particular release. I've tried to navigate the history as best I can, details below. I think they are correct.

---

The two PRs #15 and #17 are branched from a different base, and:

• they contain different versions of RunII_Autumn18_LowPtElectrons_mva_id.xml.gz
• while LowPtElectrons_ID_2020Sept15.root is identical

1. PR LowPtElectrons: latest BDT model for ID (ROOT file format) #15 is based on the branch CMSBParking:from-CMSSW_10_2_15_2020Sept15

> git log --oneline RunII_Autumn18_LowPtElectrons_mva_id.xml.gz
2aa9f04 Extra variables for LowPtElectrons Id added. Depth10, no regression
c79bcd5 added Autumn18 trainings

> git diff 2aa9f04^ 2aa9f04
diff --git a/LowPtElectrons/RunII_Autumn18_LowPtElectrons_mva_id.xml.gz b/LowPtElectrons/RunII_Autumn18_LowPtElectrons_mva_id.xml.gz
index 5c15cb1..29ce4ac 100644

Here is the diff on GitHub: CMSBParking@2aa9f04#diff-72d9ce87b279f68129c5861a9448a971a70c5b38c07373005980eadf0a990d8b.

At the head of this branch, the size of RunII_Autumn18_LowPtElectrons_mva_id.xml.gz is 8519842 bytes.

2. PT LowPtElectrons: ID models for 10_6_X #17 (this PR) is based on the branch CMSBParking:LowPtElectrons_ID_2020Sept15_for106X

> git log --oneline RunII_Autumn18_LowPtElectrons_mva_id.xml.gz gives:
c79bcd5 added Autumn18 trainings

At the head of this branch, the size of RunII_Autumn18_LowPtElectrons_mva_id.xml.gz is 7599273 bytes.

[slava77]

I still could not follow the description given what's showing up in the web interface: only LowPtElectrons/LowPtElectrons_ID_2020Sept15.root is changed in this PR, while the description is about RunII_Autumn18_LowPtElectrons_mva_id.xml.gz

[bainbrid]

This is the commit history for the branch for this PR (#17):
https://github.com/CMSBParking/RecoEgamma-ElectronIdentification/commits/from-CMSSW_10_2_15_2020Sept15

• The penultimate commit is a merge from PR Final BDT models based on 10.2 MC samples  #13 (dated Feb 2019), which contains the xml.gz file used throughout the 10_6_X cycle. Importantly, this is the file we need to maintain backward compatibility for the RECO sequence in 10_6_X.
• The last commit is the one I've just made, which adds the 2020Sept15 model for reminiAOD.

This is the commit history for the branch for PR #15:
https://github.com/CMSBParking/RecoEgamma-ElectronIdentification/commits/from-CMSSW_10_2_15_2020Sept15

• The penultimate commit (CMSBParking@4379aec) is a merge from PR LowPtElectrons: updated BDT model for ID #14 (dated Aug 2020), which contains an update to the xml.gz file, which differs from the model throughout the 10_6_X cycle. Importantly, this is the file we need to revert to maintain backward compatibility for the RECO sequence in 10_6_X.
• The last commit adds the 2020Sept15 model, which was recently merged for Add low-pT electrons to MINIAOD, update ID, improve end user experience cms-sw/cmssw#31220

This is the commit history for the master of cms-data:
https://github.com/cms-data/RecoEgamma-ElectronIdentification/commits/master

• In short, we want to revert the change in commit 2aa9f04

[slava77]

This is the commit history for the branch for this PR (#17):

all of this apparently means that we

• need a PR to a branch corresponding to 10_6_X, which does not exist
• or we need to explicitly revert the update in the xml file in the master (remake this PR on top of the master)

[slava77]

@smuzaffar @mrodozov
we will need a separate branch for this external for the update to 10_6_X.

I see that the current version in 10_6_X is V01-01-03,
perhaps V01-01-03_X would work, but feel free to suggest another branch name (and go ahead to make it, please).

[smuzaffar]

@slava77 , I created a new branch V01-01-XX (based on tag V01-01-03) and updated this PR to be based on it.
If this PR is good to go then we can merge and create a tag V01-01-04.

[slava77]

@slava77 , I created a new branch V01-01-XX (based on tag V01-01-03) and updated this PR to be based on it.
If this PR is good to go then we can merge and create a tag V01-01-04.

Thank you.
Let me see what comes out from tests with cms-sw/cmssw#32372

[slava77]

+1

@slava77 , I created a new branch V01-01-XX (based on tag V01-01-03) and updated this PR to be based on it.
If this PR is good to go then we can merge and create a tag V01-01-04.

Thank you.
Let me see what comes out from tests with cms-sw/cmssw#32372

I think that the last set of tests was good and we can go ahead with merging this PR.

[mrodozov]

+externals

[cmsbuild]

This pull request is fully signed and it will be integrated in one of the next V01-01-XX IBs after it passes the integration tests. This pull request will now be reviewed by the release team before it's merged. @silviodonato, @dpiparo, @qliphy (and backports should be raised in the release meeting by the corresponding L2)

[cmsbuild]

+1
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5d9877/11433/summary.html
CMSSW: CMSSW_11_3_X_2020-12-08-1100
SCRAM_ARCH: slc7_amd64_gcc900

[cmsbuild]

Comparison results are now available
https://cmssdt.cern.ch/SDT/jenkins-artifacts/pull-request-integration/PR-5d9877/11433/summary.html
CMSSW: CMSSW_11_3_X_2020-12-08-1100
SCRAM_ARCH: slc7_amd64_gcc900

Comparison Summary:

• No significant changes to the logs found
• Reco comparison results: 0 differences found in the comparisons
• DQMHistoTests: Total files compared: 36
• DQMHistoTests: Total histograms compared: 2747014
• DQMHistoTests: Total failures: 1
• DQMHistoTests: Total nulls: 0
• DQMHistoTests: Total successes: 2746991
• DQMHistoTests: Total skipped: 22
• DQMHistoTests: Total Missing objects: 0
• DQMHistoSizes: Histogram memory added: 0.0 KiB( 35 files compared)
• Checked 153 log files, 37 edm output root files, 36 DQM output files