Skip to content

Work In Progress: Update notebooks - #234

Open
prayagyadav wants to merge 10 commits into
iris-hep:mainfrom
prayagyadav:main
Open

Work In Progress: Update notebooks#234
prayagyadav wants to merge 10 commits into
iris-hep:mainfrom
prayagyadav:main

Conversation

@prayagyadav

@prayagyadav prayagyadav commented Jun 9, 2026

Copy link
Copy Markdown

Continuation of @oshadura 's work #232 and @maxgalli 's work for calver coffea

  • Fixed order of args for coffea runner
    Added a default value for uproot_options

  • Fixed ServiceX query to make sure it produces NanoAODSchema compatible outputs
    Added the eventID branches run, luminosityBlock and event - these are needed for the NanoAODSchema building

  • Fixed Tri-Jet 4 vector addition issue
    The ServiceX query does not output the charge for jets (and maybe jet.charge is not available for some root files?)
    When we add jet objects using the '+' operator - it's corresponding custom definition is invoked (Jet is a 'Candidate' class that adds charges too when using the add operation; If charge is not available, we see an error
    Rebuilt Jet Lorentz vector without the charge so that charge is no longer needed to add them

  • Add statistical inference notebooks and datacard from calvert coffee and save the root files such that they are compatible with datacard.
    Added an extra cell in the ttbar-pipeline notebook: To save the histograms and rootfiles in a manner that is compatible with the datacard and combine in general
    A few things changed:
    Done i) Rebin histograms before saving into the root files
    Done ii) Rename all _up/_down to Up/Down
    Done iii) Rename all scaleup/scaleDown to scaleUp/scaleDown
    Done iv) Rename pseudodata -> data_obs (Probably unnecessary?)
    Done v) Copy the symmetric histograms to save both Up and Down variation separately
    Done vi) Save the histogram in root files that are named appropriately with respect to the datacard being used

  • in training notebook was fixed scatter large arrays before client.map to avoid the Dask 2025.3.1 scheduler bug

  • fix the issue that scikit-learn 1.8 now returns Python native float from metrics like accuracy_score, whereas older versions returned numpy.float64, which has a .round() method

@oshadura
oshadura requested a review from alexander-held June 9, 2026 08:32
@oshadura

Copy link
Copy Markdown
Member

@prayagyadav this is still work in progress? Or we can invite reviewers to take a look?

@prayagyadav

Copy link
Copy Markdown
Author

@prayagyadav this is still work in progress? Or we can invite reviewers to take a look?

Sorry @oshadura I am still working on it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants