Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-11962

Finalize Timeseries features to be computed on DIAObjects

    Details

      Description

      The DPDD specifies that DIAObjects will have periodic and non-periodic timeseries features computed on them. These are baselined to be those in Richards et al. 2011, but the DPDD is clear that the choice of features should be re-examined closer to operations to incorporate new developments in the field.

      This ticket is to update the DPDD with new features developed in consultation with the science collaborations.

        Attachments

          Issue Links

            Activity

            Hide
            ebellm Eric Bellm added a comment -

            This page https://www.kaggle.com/michaelapers/the-plasticc-astronomy-classification-demo provides a useful list of existing relevant classification work as well as feature construction libraries. It strikes me that the output of PLAsTiCC should be useful in guiding this study.

            Show
            ebellm Eric Bellm added a comment - This page https://www.kaggle.com/michaelapers/the-plasticc-astronomy-classification-demo provides a useful list of existing relevant classification work as well as feature construction libraries. It strikes me that the output of PLAsTiCC should be useful in guiding this study.
            Hide
            ebellm Eric Bellm added a comment -

            The TVS Task Force has apparently become active but is approaching the problem by considering a small handful of object types, which I do not think is the best approach.

            I note that the winning entry to PLAsTiCC used GPR to flexibly parameterize lightcurve behavior (e.g., the existing and measurement of peaks) in a way that may be useful when considering non-periodic features for transients.

            I note that there may be some tension here between providing precomputed features that are valuable as inputs to machine-learned models, and features that may be more interpretable and hence useful for human querying and filtering of DIAObjects.

            Show
            ebellm Eric Bellm added a comment - The TVS Task Force has apparently become active but is approaching the problem by considering a small handful of object types, which I do not think is the best approach. I note that the winning entry to PLAsTiCC used GPR to flexibly parameterize lightcurve behavior (e.g., the existing and measurement of peaks) in a way that may be useful when considering non-periodic features for transients. I note that there may be some tension here between providing precomputed features that are valuable as inputs to machine-learned models, and features that may be more interpretable and hence useful for human querying and filtering of DIAObjects.
            Hide
            ebellm Eric Bellm added a comment -

            (continuing to capture notes here: some transient-specific features in http://adsabs.harvard.edu/abs/2019arXiv190400014M)

            Show
            ebellm Eric Bellm added a comment - (continuing to capture notes here: some transient-specific features in http://adsabs.harvard.edu/abs/2019arXiv190400014M )
            Hide
            ebellm Eric Bellm added a comment -

            Noting here because I didn't realize it earlier: LDM-151 section 6.21 has an enumerated set of baselined periodic and non-periodic features (again, derived from Butler et al. 2011).

            Show
            ebellm Eric Bellm added a comment - Noting here because I didn't realize it earlier: LDM-151 section 6.21 has an enumerated set of baselined periodic and non-periodic features (again, derived from Butler et al. 2011).
            Hide
            mgraham Melissa Graham added a comment -

            There's no way you could have known this but, FYI: the attached ascii file MLG_email_TVS_VarChar.txt has a list of everywhere I could find that DM documentation has content related to variability parameters. 

            Show
            mgraham Melissa Graham added a comment - There's no way you could have known this but, FYI: the attached ascii file MLG_email_TVS_VarChar.txt has a list of everywhere I could find that DM documentation has content related to variability parameters. 

              People

              • Assignee:
                ebellm Eric Bellm
                Reporter:
                ebellm Eric Bellm
                Watchers:
                Eric Bellm, Leanne Guy, Melissa Graham
              • Votes:
                0 Vote for this issue
                Watchers:
                3 Start watching this issue

                Dates

                • Due:
                  Created:
                  Updated:

                  Summary Panel