Results.md

Planet Hunters TESS Eclipsing Binary Candidate Catalog

At a glance

Num. of PHT Subjects	28092
Num. of EBs / EB Candidates (TICs)	12561
Num. of EBs / EB Candidates with high certainty	4066
Estimated num. of new EBs	1480
Sectors	1 - 39
Num. of users	1320

The candidates are cross-matched against SIMBAD, VSX, and ASAS-SN to

• differentiate between known eclipsing binary and candidates .
• calculate a proxy of tagging accuracy.

Num. of EBs / EB Candidates with high certainty: defined by eb_score >= 3, which roughly means at least 3 users have tagged a subject as an eclipsing binary. The subset has a proxy accuracy >= 90%.

See:

• dashboard notebook for more detailed breakdown.
• the catalog in csv
• the pilot study. The current result uses similar methodology, and applying it to the complete data set of sectors 1 - 39.
  • Note: the eb_score used here corresponds to N_eb_adj used in the pilot study.

User Contributions

• 1320 users contributed to the tagging.
• The proxy tagging accuracy remains quite stable over the 3 year period: ~90+% ateb_score >= 3, or ~80% overall.
  • In fact, there was some decrease in sectors 30 -39
• Top users contributed to majority of the tagging:

Rank	Cumulative Percentage
1	39%
5	52%
20	64%
50	73%
100	81%

• Some form of user weighting could be helpful.

See participants dashboard notebook for more detailed breakdown.

Next Steps

• Vetting of a subset of targets to answer
  • for those counted as false positives (is_eb_catalog == "F"), how many are indeed false positives? Is there some pattern? Some suspicions include:
    • Those listed as RR Lyrae in other catalogs are probably genuine false positives: users probably mistreat them as w-uma
    • Those listed as rotators in other catalogs: some of them possibly have real eclipses in addition to the rotational variability listed, and should be counted as proper match.
  • for those counted as no data in other catalogs (is_eb_catalog == "-")
    • what is the tagging accuracy?
    • A number of classifications are treated as no data as they are deemed to be irrelevant in the context of eclipsing binary, e.g., star in SIMBAD, various eruptive / cataclysmic types. Is such treatment appropriate?
• Review cross matching (with SIMBAD, etc.) to see if we have included too many invalid ones (false positives) or excluded too many genuine ones (false negatives)
  • When matching plx / PM, consider to use the error supplied in the catalog to determine if it is a match.
• Review tag tallying and computation of eb_score
  • Add tags to be counted, in particular, #EB and #E.B. are employed by some users, but they are not treated as tags in Zooniverse.
    • There are ~3000 such comments, while there are about ~60000 eclipsingbinary comments.
    • other tags to consider: #possibleEB, #eclipsing_binary, #eclipsing_binaries, #eclipsingbinaries, #eclipsing-binaries
  • Handle cases that users tag a subject both as eclipsing binary and transit. (Currently it is treated as a neutral vote)
  • Consider additional tags counted as dissenting voices. Candidates include #rr-lyrae (and possibly other pulsators / rotators)
  • Should we consider #contamination and/or #NEB ?
• Cross match with Gaia DR3 for variable status, RUWE, etc.
• Produce a list of vetted candidates.

Data Sources / Credits

• Planet Hunters TESS participants
• The 3 catalogs: VSX, ASAS-SN, SIMBAD
  • Access is done by using astroquery package via Vizier.
• TIC parameters from MAST
  • TIC parameters access is done by using astroquery.