J/A+A/616/A97ivo://CDS.VizieR/J/A+A/616/A97doi:10.26093/cds/vizier.36160097CDSD'Isanto A.Cavuoti S.Gieseke F.Polsterer K.L.2018-09-27T12:13:21Z2018-08-28T08:50:25ZCDS support team

CDS, Observatoire de Strasbourg, 11 rue de l'Universite, F-67000 Strasbourg, France

cds-question@unistra.frquasarssurveysThe explosion of data in recent years has generated an increasing need for new analysis techniques in order to extract knowledge from massive data-sets. Machine learning has proved particularly useful to perform this task. Fully automatized methods (e.g. deep neural networks) have recently gathered great popularity, even though those methods often lack physical interpretability. In contrast, feature based approaches can provide both well-performing models and understandable causalities with respect to the correlations found between features and physical processes. Efficient feature selection is an essential tool to boost the performance of machine learning models. In this work, we propose a forward selection method in order to compute, evaluate, and characterize better performing features for regression and classification problems. Given the importance of photometric redshift estimation, we adopt it as our case study. We synthetically created 4520 features by combining magnitudes, errors, radii, and ellipticities of quasars, taken from the Sloan Digital Sky Survey (SDSS). We apply a forward selection process, a recursive method in which a huge number of feature sets is tested through a k-Nearest-Neighbours algorithm, leading to a tree of feature sets. The branches of the feature tree are then used to perform experiments with the random forest, in order to validate the best set with an alternative model. We demonstrate that the sets of features determined with our approach improve the performances of the regression models significantly when compared to the performance of the classic features from the literature. The found features are unexpected and surprising, being very different from the classic features. Therefore, a method to interpret some of the found features in a physical context is presented. The feature selection methodology described here is very general and can be used to improve the performance of machine learning models for any regression or classification task.2018A&A...616A..97Dhttps://cdsarc.cds.unistra.fr/viz-bin/cat/J/A+A/616/A97CatalogResearchIsServedByTAP VizieR generic serviceIsServedByConesearch servicerelated-toII/294 : The SDSS Photometric Catalog, Release 7 (Adelman-McCarthy+, 2009)V/139 : The SDSS Photometric Catalog, Release 9 (Adelman-McCarthy+, 2012)VII/260 : The SDSS-DR7 quasar catalog, DR7Q (Schneider+, 2010)VII/269 : The SDSS-DR9 quasar catalog, DR9Q (Paris+, 2012)http://www.sdss.org/dr7 : SDSS DR7 home pagehttp://www.sdss3.org/dr9 : SDSS DR9 home pagehttps://cds.unistra.fr/vizier-org/licences_vizier.htmlhttps://vizier.cds.unistra.fr/viz-bin/VizieR-2?-source=J/A+A/616/A97https://vizier.iucaa.in/viz-bin/VizieR-2?-source=J/A+A/616/A97http://vizieridia.saao.ac.za/viz-bin/VizieR-2?-source=J/A+A/616/A97https://vizier.cds.unistra.fr/viz-bin/votable?-source=J/A+A/616/A97https://vizier.iucaa.in/viz-bin/votable?-source=J/A+A/616/A97http://vizieridia.saao.ac.za/viz-bin/votable?-source=J/A+A/616/A97GETtext/xml+votablehttps://tapvizier.cds.unistra.fr/TAPVizieR/tapCone search capability for table J/A+A/616/A97/dr7_9 (SDSS object IDs and coordinates of the quasars for experiment DR7+9)https://vizier.cds.unistra.fr/viz-bin/conesearch/J/A+A/616/A97/dr7_9?https://vizier.iucaa.in/viz-bin/conesearch/J/A+A/616/A97/dr7_9?http://vizieridia.saao.ac.za/viz-bin/conesearch/J/A+A/616/A97/dr7_9?GETtext/xml+votable180.050000true0.0272280.5153410.005555555555555556Cone search capability for table J/A+A/616/A97/dr7a (SDSS object IDs and coordinates of the quasars for experiment DR7a)https://vizier.cds.unistra.fr/viz-bin/conesearch/J/A+A/616/A97/dr7a?https://vizier.iucaa.in/viz-bin/conesearch/J/A+A/616/A97/dr7a?http://vizieridia.saao.ac.za/viz-bin/conesearch/J/A+A/616/A97/dr7a?GETtext/xml+votable180.050000true146.43803677-0.656042390.005555555555555556Cone search capability for table J/A+A/616/A97/dr7b (SDSS object IDs and coordinates of the quasars for experiment DR7b)https://vizier.cds.unistra.fr/viz-bin/conesearch/J/A+A/616/A97/dr7b?https://vizier.iucaa.in/viz-bin/conesearch/J/A+A/616/A97/dr7b?http://vizieridia.saao.ac.za/viz-bin/conesearch/J/A+A/616/A97/dr7b?GETtext/xml+votable180.050000true118.1953161838.961204090.0055555555555555561/5 8 10 27 2/16-17 19 25 28 36 38 103 107 3/72-73 75 97 116 176-178 184 407 411 423 427 494 4/296-297 299 399 468-469 472 592 594 600 602 624 716 718 740 744-746 1367 1375 1397 1621-1623 1625-1627 1637-1639 1641-1643 1685-1687 1690-1691 1701-1703 1705-1707 1961-1963 1965-1967 1980 1982-1983 2026-2027 2031 5/7 12 136 138 178 425 518 523 551 574 1192-1193 1195 1540-1541 1549 1585 1588-1589 1684-1685 1728-1729 1732-1733 1790-1791 1880-1882 1896-1897 1900-1901 2041 2372 2374-2375 2380-2381 2404 2504 2506 2518 2521 2523-2524 2526 2528 2530 2536 2545 2547-2548 2550 2553 2555 2868 2870 2876 2878 2897-2899 2968 2970 2988 3077 3079 3088 3095 3097-3100 3136-3138 3140 3332 4303 4307-4310 4312-4314 4323-4326 4329-4330 4351 4373-4375 4377-4380 4387 4389-4390 4393-4394 4416 4435-4438 4441-4442 4444 4451-4454 4457-4458 4460 4475-4476 4478 4501-4502 4505-4507 4516-4518 4521-4523 4544 4551 4557-4558 4560-4562 4575 4596-4598 4601-4602 4629-4630 4633-4636 4645 4647 4651-4652 4656 4693-4698 4709-4710 4713-4714 4736-4738 4756-4758 4761-4762 4764 4772-4774 4777-4778 4780 4791 4793-4796 4817 4831-4833 4845-4847 4851-4854 4856 4864 4866 4872 4899 4917-4919 4921-4924 4933-4935 4937-4940 4960 4985 4997-4998 5000-5002 5042 5118 5445 5461-5463 5466 5491-5493 5596-5597 5599 5621 5623 5679 5761-5764 5768 5786 5797-5799 5801-5802 5835 6397-6399 6461-6463 6477-6479 6481-6483 6497-6499 6541-6543 6545-6547 6561-6563 6717-6719 6737-6739 6754-6755 6757-6759 6797-6799 6801-6803 6817 6819 7509-7510 7823 7829-7830 7833-7834 7839 7841 7843 7856 7858-7859 7882-7883 7886 7924 7926 8098-8099 8102-8103 8114 8119-8120 8122 8168 8170-8171 8174 9179 9181-9183 9191 9193-9198 9200-9202 9214-9215 11261-11263 12247 12252-12254 12271 12273-12276 12280 6/0-2 4 19 22-23 25-27 37-38 41-42 57 64 72 128 260 521 523-524 526-527 548-550 556 640 706-707 716 718 726-727 733 736-737 896 898-899 902 904 907-910 914 920 971-972 975 991-993 995 997 1009 1014 1678 1697 1705 1712 2023 2032-2033 2040 2053 2055 2061 2064 2066-2068 2070 2076 2086 2089 2091 2177 2180 2198-2199 2201 2228-2229 2248 2250 2285 2287 2290 2300 2302 2640-2641 2644 3965 4053 4777 4779 6148-6149 6151 6168-6169 6172-6173 6175 6204-6205 6207 6295 6301-6303 6306 6312 6314 6336-6337 6339 6344-6345 6351 6362-6363 6365-6367 6372-6373 6375 6381 6705-6706 6708 6710 6712 6723 6734 6976-6977 6980-6981 6992-6993 6996-6997 7159 7532-7533 7536-7538 7540 7568-7569 7572-7573 7582-7583 7592-7593 7595-7597 7608-7609 7612 7680-7681 7684-7685 7696-7697 7700 7779 7782 7784 7787 7842 7848 8126-8127 8139 8141-8143 8163 8170 9492 9494 9528-9530 9532 9535 9568 9570 9576 9578 9621 9623-9624 9626 9629 9648 9650 9656 9658 10020-10022 10028 10030 10061-10063 10067 10116 10118 10124 10152-10154 10177 10179 10185 10187 10209 10211 10217 10219 10224-10226 10232 10234-10235 10238 11476-11478 11484 11486 11508 11520 11522 11528 11530 11585-11587 11600-11602 11604 11608 11610 11616-11617 11856 11858 11864 11876-11878 11884 11956 11960 11968-11970 12288-12290 12305 12307 12313 12356 12360 12374-12375 12377-12379 12387 12404-12406 12408-12410 12480-12481 12484 12519 12523 12536 12538 12556-12557 12564 12576 12658-12660 12662-12663 12665 12805 12818 12820 12822 12865 13317 13332 13413 13420-13421 13426 13432 13637 13639 13648-13650 13657-13658 13660 13680-13682 13689-13691 13765 13767 13773 13786 13799 13808 13810 14586 14910 14921 14993 14996 14998-14999 15911 15917-15919 15928-15929 16005 16010 16185 16187 16273-16274 16278 16291 16295 16299 16301 16799 16887 16891 16893-16895 17149-17151 17207 17211 17223 17244 17246 17260-17262 17264 17287 17291 17308-17309 17313-17315 17324-17326 17328-17330 17344-17346 17357 17359 17362 17368-17370 17393-17396 17398-17401 17403 17469-17471 17485-17487 17491 17505-17507 17524 17547 17553-17555 17564-17566 17569-17571 17580-17582 17584-17586 17600-17602 17604 17655 17659 17661-17663 17668-17669 17672-17673 17680-17681 17723 17725-17729 17732-17733 17756-17758 17761-17763 17772-17774 17784 17803 17820-17822 17825-17827 17836-17838 17848 17856-17858 17860 17877-17879 17883 17895 17899 17908 17910-17911 17916 17920-17921 17981-17983 17997-17999 18001-18003 18012-18014 18017-18019 18032-18034 18040 18061-18063 18076-18078 18081-18083 18096-18098 18112-18114 18157 18159 18168 18171 18199 18221 18223 18225-18227 18236-18238 18252-18254 18256 18258 18272-18274 18293-18295 18299 18317-18319 18321 18328 18365-18367 18375 18381-18383 18396 18398 18401-18403 18412-18414 18416-18418 18495 18509-18511 18514-18515 18524-18526 18531 18548 18552 18585-18587 18597-18599 18601 18603 18612-18614 18616-18618 18628 18630 18632 18634 18685-18690 18749-18751 18763 18765-18767 18769-18771 18796-18798 18800-18802 18804-18805 18807-18808 18829-18831 18833-18835 18844-18846 18849-18851 18860-18862 18864-18866 18880-18882 18927 18956 18960 18976 18978 18989 18991 19003 19005-19007 19021-19023 19036-19038 19041-19043 19052-19054 19085-19087 19100-19102 19105-19107 19116-19118 19135-19138 19144 19151 19157-19159 19161-19163 19171 19188-19190 19192-19194 19200-19202 19238-19239 19241 19243-19244 19280 19319 19322-19323 19336-19338 19340 19344 19360 19374-19375 19379 19397-19399 19401-19403 19420 19428-19430 19432-19433 19494 19496-19497 19500 19525 19536-19538 19540 19583 19590 19608 19610 19621 19646-19647 19663 19665 19667 19681 19683 19700 19704 19725-19727 19729-19731 19745 19747 19764 19768 19776-19778 19780-19781 19844 19918-19919 19930 19949 19952 19954 19960 19979 19981-19983 19985-19987 19996-19998 20012-20014 20016-20018 20032-20034 20036 20096 20143 20154-20155 20158 20162-20163 20172 20193 20239 20243 20250-20252 20254 20272-20273 20277-20279 20284-20285 20287 20328 20452-20454 20457 20460-20461 20464 20466 20478 21769 21771 21823 21839-21841 21843 21857 21859 21861-21863 21868-21869 21909-21910 21952 21959 21976 21978-21979 21981-21983 21991 21997-21999 22341 22349 22351 22373 22393 22395 22451 22462 22481 22517 22519 22650-22651 22654-22655 22711 22714-22715 22736-22737 22739 22742-22743 22760 22762-22763 22887 22892 22898 23043 23060-23062 23064-23066 23076-23077 23080 23099 23101-23104 23138-23139 23150 23181-23183 23185-23187 23201-23203 23212-23214 23216-23218 23231-23234 23240 23272 23274 23278 23352 23354-23355 23405-23406 23416 23418 23428-23429 23470 24053 24055 24057 24060-24061 24237 24251 25247 25289-25291 25294 25312-25313 25316 25331-25332 25335 25340 25517 25519 25528 25530-25531 25565-25567 25581-25583 25585-25587 25771 25821-25823 25837-25839 25841-25843 25885-25887 25901-25903 25905-25907 25923 25985-25987 26112-26113 26115-26116 26118 26121 26141-26143 26157-26159 26161-26163 26177-26179 26241-26243 26846-26847 26862-26863 26865 26867 26885 26909-26911 26933 26935 26941-26943 26945-26947 27009-27012 27014-27015 27025-27027 27167 27183 27187 27201 27203 27267 27273-27275 29777 29973 30029-30031 30033-30035 30044-30046 30053 30064-30066 30068 30975 31139 31145-31147 31289-31291 31309 31311 31315 31324 31330-31331 31340 31344 31351 31353-31355 31369-31371 31429-31431 31488 31522 31550-31551 31594 31598-31599 31611 31700 31710 31880-31881 31886 31905 31908 32180 32386 32390-32391 32402-32403 32406-32407 32451 32460 32462-32463 32473-32475 32484 32486 32492 32494-32495 32519 32530 32537 32553 32555-32556 32558 32641-32642 32648 32650-32651 32678-32679 32690 32702 36607 36702-36703 36715 36721-36723 36759 36761-36763 36796 36798-36799 36812 36816-36818 36832 36842 36845-36847 36850-36851 40949-40951 40957-40959 45023 45039 45043 48894 48983 48987 49005 49007 49020-49022 49077-49079 49091 49108 49112 49124 49126 49128-49130 49149-49151https://cdsarc.cds.unistra.fr/viz-bin/moc/J/A+A/616/A97?format=asciidefaultJ/A+A/616/A97/dr7_9SDSS object IDs and coordinates of the quasars for experiment DR7+9recnoRecord number assigned by the VizieR team. Should Not be used for identification.meta.recordintobjIDSDSS DR7 objIDmeta.idcharRAJ2000Right ascension (J2000.0)degpos.eq.ra;meta.mainindexedDEJ2000Declination (J2000.0)degpos.eq.dec;meta.mainindexedSDSS-DR7Display the SDSS data for this objectmeta.ref.urlcharJ/A+A/616/A97/dr7aSDSS object IDs and coordinates of the quasars for experiment DR7arecnoRecord number assigned by the VizieR team. Should Not be used for identification.meta.recordintobjIDSDSS DR7 objIDmeta.idcharRAJ2000Right ascension (J2000.0)degpos.eq.ra;meta.mainindexedDEJ2000Declination (J2000.0)degpos.eq.dec;meta.mainindexedSDSS-DR7Display the SDSS data for this objectmeta.ref.urlcharJ/A+A/616/A97/dr7bSDSS object IDs and coordinates of the quasars for experiment DR7brecnoRecord number assigned by the VizieR team. Should Not be used for identification.meta.recordintobjIDSDSS DR7 objIDmeta.idcharRAJ2000Right ascension (J2000.0)degpos.eq.ra;meta.mainindexedDEJ2000Declination (J2000.0)degpos.eq.dec;meta.mainindexedSDSS-DR7Display the SDSS data for this objectmeta.ref.urlchar