Hi, can one assume that there are no catastrophic spectroscopic/photometric redshift errors due to attributing objects to the wrong host galaxy or should this be considered a possibility (e.g. as per https://arxiv.org/abs/1604.06138)?
", "messages": [ { "id": "396965", "postDate": "10/01/2018 17:35:29", "content": "Hi, can one assume that there are no catastrophic spectroscopic/photometric redshift errors due to attributing objects to the wrong host galaxy or should this be considered a possibility (e.g. as per https://arxiv.org/abs/1604.06138)?
", "rawMarkdown": "Hi, can one assume that there are no catastrophic spectroscopic/photometric redshift errors due to attributing objects to the wrong host galaxy or should this be considered a possibility (e.g. as per https://arxiv.org/abs/1604.06138)?", "votes": null }, { "id": "396991", "postDate": "10/01/2018 18:13:52", "content": "Per page 8 of the data note:
\n\n\n", "rawMarkdown": "Per page 8 of the data note:\n\n> Beware that a few percent of the test data redshifts are catastrophic,\n> meaning that some redshift uncertainties greatly underestimate the\n> difference between measured and true redshift.", "votes": null }, { "id": "397041", "postDate": "10/01/2018 19:48:38", "content": "Beware that a few percent of the test data redshifts are catastrophic,\n meaning that some redshift uncertainties greatly underestimate the\n difference between measured and true redshift.
\n
Thanks for the clarification, though this doesn't fully answer my question as even if there is no host misidentification, photometric redshifts will typically exhibit some catastrophic errors.
", "rawMarkdown": "Thanks for the clarification, though this doesn't fully answer my question as even if there is no host misidentification, photometric redshifts will typically exhibit some catastrophic errors.", "votes": null }, { "id": "397147", "postDate": "10/02/2018 01:24:01", "content": "All we’re prepared to say is that different effects may contribute to the bias between truth, and either spectroscopic or photometric redshift.
\n\nThat said... we're interested in classification based on the light curves. This focus helps balance the competition and limit the advantages astrophysicists engaged in the contest have. We’ve only provided minimal contextual information for each object - anything described in the starter kit/data note. In particular, none of the other features in Table 3 of the Gupta et al, '16 are included.
\n\nSo while we can't specify if host-misidentification is included, we can say you don't have the contextual information to correct for it using the method described in that paper.
\n\nIn general, if you can't expect a member of the public without a background in astrophysics to do something with the data, then you probably don't have to it!
\n\nThis still doesn't answer your question, but its the best we can do!
\n\n-Gautham for the PLAsTiCC team
", "rawMarkdown": "All we’re prepared to say is that different effects may contribute to the bias between truth, and either spectroscopic or photometric redshift. \n\nThat said... we're interested in classification based on the light curves. This focus helps balance the competition and limit the advantages astrophysicists engaged in the contest have. We’ve only provided minimal contextual information for each object - anything described in the starter kit/data note. In particular, none of the other features in Table 3 of the Gupta et al, '16 are included. \n\nSo while we can't specify if host-misidentification is included, we can say you don't have the contextual information to correct for it using the method described in that paper. \n\nIn general, if you can't expect a member of the public without a background in astrophysics to do something with the data, then you probably don't have to it! \n\nThis still doesn't answer your question, but its the best we can do!\n\n-Gautham for the PLAsTiCC team", "votes": null }, { "id": "397422", "postDate": "10/02/2018 13:13:28", "content": "Thanks, interesting issues to reflect on.
", "rawMarkdown": "Thanks, interesting issues to reflect on.", "votes": null } ], "comments": [ { "id": 396991, "author_name": "sohier", "author_url": "", "post_date": "10/01/2018 18:13:52", "content": "Per page 8 of the data note:
\n\n\n", "votes": null, "replies": [ { "id": 397041, "author_name": "brucebassett", "author_url": "", "post_date": "10/01/2018 19:48:38", "content": "Beware that a few percent of the test data redshifts are catastrophic,\n meaning that some redshift uncertainties greatly underestimate the\n difference between measured and true redshift.
\n
Thanks for the clarification, though this doesn't fully answer my question as even if there is no host misidentification, photometric redshifts will typically exhibit some catastrophic errors.
", "votes": null, "replies": [] }, { "id": 397147, "author_name": "gsnarayan", "author_url": "", "post_date": "10/02/2018 01:24:01", "content": "All we’re prepared to say is that different effects may contribute to the bias between truth, and either spectroscopic or photometric redshift.
\n\nThat said... we're interested in classification based on the light curves. This focus helps balance the competition and limit the advantages astrophysicists engaged in the contest have. We’ve only provided minimal contextual information for each object - anything described in the starter kit/data note. In particular, none of the other features in Table 3 of the Gupta et al, '16 are included.
\n\nSo while we can't specify if host-misidentification is included, we can say you don't have the contextual information to correct for it using the method described in that paper.
\n\nIn general, if you can't expect a member of the public without a background in astrophysics to do something with the data, then you probably don't have to it!
\n\nThis still doesn't answer your question, but its the best we can do!
\n\n-Gautham for the PLAsTiCC team
", "votes": null, "replies": [] }, { "id": 397422, "author_name": "brucebassett", "author_url": "", "post_date": "10/02/2018 13:13:28", "content": "Thanks, interesting issues to reflect on.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "396965": "Hi, can one assume that there are no catastrophic spectroscopic/photometric redshift errors due to attributing objects to the wrong host galaxy or should this be considered a possibility (e.g. as per https://arxiv.org/abs/1604.06138)?", "396991": "Per page 8 of the data note:\n\n> Beware that a few percent of the test data redshifts are catastrophic,\n> meaning that some redshift uncertainties greatly underestimate the\n> difference between measured and true redshift.", "397041": "Thanks for the clarification, though this doesn't fully answer my question as even if there is no host misidentification, photometric redshifts will typically exhibit some catastrophic errors.", "397147": "All we’re prepared to say is that different effects may contribute to the bias between truth, and either spectroscopic or photometric redshift. \n\nThat said... we're interested in classification based on the light curves. This focus helps balance the competition and limit the advantages astrophysicists engaged in the contest have. We’ve only provided minimal contextual information for each object - anything described in the starter kit/data note. In particular, none of the other features in Table 3 of the Gupta et al, '16 are included. \n\nSo while we can't specify if host-misidentification is included, we can say you don't have the contextual information to correct for it using the method described in that paper. \n\nIn general, if you can't expect a member of the public without a background in astrophysics to do something with the data, then you probably don't have to it! \n\nThis still doesn't answer your question, but its the best we can do!\n\n-Gautham for the PLAsTiCC team", "397422": "Thanks, interesting issues to reflect on." }, "source": "meta" }