Kubva Pakuvanzika kusvika Pazvinogona: Kushandisa Synthetic Data kuburikidza neiyo yakabatanidzwa Syntho Injini muSAS Viya sechikamu cheSAS Hackathon kuvhura zvakavanzika data.

Isu tinovhura iyo yakazara mukana wedata rehutano neino gadzira AI panguva yeSAS Hackathon.

Nei uchivhurira zvakavanzika zvehutano data?

Hutano hunoda zvakanyanya data drive ruzivo. Nekuti hutano hwakashomeka vashandi, vakamanikidzwa zvakanyanya nekugona kuponesa hupenyu. Nekudaro, data rehutano ndiyo yakanyanya kuvanzika data uye saka yakavharwa. Iyi data inobata zvakavanzika:

  • Zvinotora nguva kuwana
  • Inoda mapepa akawanda
  • Uye haigoni kushandiswa chete

Izvi zvinonetsa, sechinangwa chedu cheiyi hackathon ndiyo inofanotaura kudzikira uye kufa sechikamu chekutsvagisa cancer kuchipatara chinotungamira. Ndosaka Syntho neSAS vachibatirana kuchipatara ichi, uko Syntho anovhura data nedata rekugadzira uye SAS inoona data ruzivo neSAS Viya, inotungamira analytics chikuva.

Synthetic data?

Yedu Syntho Injini inogadzira nyowani nyowani yakagadzirwa data. Musiyano wakakosha, isu tinoshandisa AI kutevedzera hunhu hwe data renyika chaiyo mune data rekugadzira, uye kusvika padanho rekuti rinogona kutoshandiswa pakuongorora. Ndosaka tichiidaidza kuti synthetic data twin. Iyo yakanaka seiyo chaiyo uye nhamba yakafanana neiyo yekutanga data, asi pasina njodzi yekuvanzika.

Syntho Injini yakabatanidzwa muSAS Viya

Munguva iyi hackathon, takabatanidza iyo Syntho Engine API muSAS Viya senhanho. Pano takasimbisawo kuti data rekugadzira rakaita senge chairo muSAS Viya. Tisati tatanga netsvakiridzo yegomarara, takaedza nzira iyi yakasanganiswa nedataset yakavhurika uye yakasimbiswa kana iyo data yekugadzira iri chaiyo-yakanaka-sechokwadi kuburikidza nenzira dzakasiyana dzekusimbisa muSAS Viya.

Ko synthetic data seyakanaka-sechokwadi?

Iko kuwirirana, hukama pakati pezvakasiyana, zvakachengetedzwa.

Iyo Nzvimbo Pasi pe curve, chiyero chekushanda kwemuenzaniso, inochengetedzwa.

Uye kunyange kukosha kwekusiyana, simba rekufungidzira remhando yemuenzaniso, rinobata kana tichienzanisa data yepakutanga ne data yekugadzira.

Nekudaro, isu tinogona kugumisa kuti synthetic data inogadzirwa neSyntho Engine muSAS Viya ndeyechokwadi-yakanaka-se-chaiyo uye kuti isu tinogona kushandisa synthetic data yekuvandudza modhi. Nekudaro, isu tinogona kutanga nekutsvagisa kwegomarara uku kufanotaura kuderera uye kufa.

Synthetic data yeCancer Research yechipatara chinotungamira

Pano, isu takashandisa iyo yakabatanidzwa Syntho Injini senhanho muSAS Viya kuvhura iyi yakavanzika data ine synthetic data.

Mhedzisiro yacho, AUC ye0.74 uye modhi inokwanisa kufanotaura kuderera uye kufa.

Nekuda kwekushandisa data rekugadzira, takakwanisa kuvhura iyi hutano mune imwe mamiriro asina njodzi shoma, yakawanda data uye nekukurumidza kuwana data.

Sanganisa data kubva kuzvipatara zvakawanda

Izvi hazvigoneke chete mukati mechipatara, zvakare data kubva kuzvipatara zvakawanda inogona kusanganiswa. Nekudaro, danho raitevera raive rekubatanidza data kubva kuzvipatara zvakawanda. Yakasiyana data yechipatara yakagadziridzwa seyekupinza modhi muSAS Viya kuburikidza neSyntho Injini. Pano, takaona AUC ye0.78, ichiratidza kuti data rakawanda rinoguma mune zviri nani kufanotaura simba reiyo modhi.

Results

Uye izvi ndizvo mhedzisiro kubva kune iyi hackathon:

  • Syntho inosanganiswa muSAS Viya sedanho
  • synthetic data inogadzirwa zvinobudirira kuburikidza neSyntho muSAS Viya
  • Synthetic data kurongeka kunotenderwa, seMamodheru akadzidziswa pane synthetic data mamakisi akafanana uye mamodheru akadzidziswa pane yekutanga data.
  • isu takafanotaura kuderera uye kufa pane synthetic data sechikamu chekutsvagisa cancer
  • uye yakaratidza kuwedzera kweAUC kana ichibatanidza data rekugadzira kubva kuzvipatara zvakawanda.

Matanho anotevera

Matanho anotevera ndeeku

  • zvinosanganisira zvimwe zvipatara
  • kuwedzera kushandiswa kwemakesi uye
  • kuwedzera kune chero imwe sangano, sezvo matekiniki ari sector agnostic.

Iyi ndiyo nzira iyo Syntho neSAS vanovhura nayo data uye vanoona data inofambiswa ruzivo muhutano hwehutano kuve nechokwadi chehutano hunoshanda zvakanaka, nekumanikidzwa kwakajairwa kuponesa hupenyu.

Synthetic Data mune Healthcare chivharo

Sevha yako synthetic data mumushumo wehutano!