I-coefficient ye-correlation - isici se-model model

I-model yokulungiswa (CM) yinkqubo yokubala ehlinzeka ngokulingana kwemathemikhali lapho isibonakaliso esisebenzayo sinqunywe khona kuye ngesibonakaliso esisodwa noma ngaphezulu.

Yx = a0 + a1x1

Kuphi: y - inkomba ephumayo, kuye ngokuthi isici x;

I-X yisici sesici;

I-A1 yi-parameter ye-CM ekhombisa ukuthi izimpikiswano ezizayo zizoshintsha kanjani njengoba isici x sishintsha ngenye, uma kulesi simo zonke ezinye izici ezithintayo zihlala zingashintshi;

I-Ao yi-parameter ye-CM, ebonisa ithonya lazo zonke ezinye izici embonini ephumayo y, ngaphandle kwesignali yesici x

Uma ukhetha imingcele ephumelelayo neyiyoqobo yemodeli, kubalulekile ukucatshangelwa ukuthi isibonakaliso esiphumelelayo kulesi sikhungo sobudlelwano obangela imbangela sisezingeni eliphakeme kunezinkomba zezici.

Izimpawu zesimodeli sokulungisa

Ngemuva kokubala imingcele ye-model model, i-coefficient yokuhlanganiswa ibalwa.

I-coefficient ye-pair correlation, -1 ≤ p ≤ 1, ibonisa amandla nesiqondiso sethonya le factor facult on the resultant. Kusondele kokungu-1, ukuxhumeka okunamandla, okuseduze no-0, kunamandla okuxhumana. Uma i-coefficient yokuhlanganiswa inenani elihle, ukuxhumeka kuqondile, uma okungalungile kuyinhloko.

I-coefficient yokuhlanganiswa kwefomula: pxy = (xy-x * 1 / y) / ey ey

I-Ex = xx2- (x) 2; Yy = y2- (y) 2

Uma i-CM yinkimbinkimbi ehlukahlukene, enefomu:

I-Yx = a0 + a1x1 + a2x2 + ... + i-anxn

Khona-ke u-coefficient multiple correlation ubalwa kuwo.

0 ≤ Р ≤ 1 futhi ikhombisa amandla ethonya lazo zonke izinto ezihlangene ndawonye ekuphumeni.

P = 1- ((yx-yu) 2 / (yu-oo) 2)

Kuphi: yh - inkomba ephumela - inani elibalwe;

Yi-inani langempela;

Inani langempela, elilinganiselwe.

Inani elibalwa le-yx litholakala ngenxa yokufaka esikhundleni esimisweni sokulungiswa kwe-x1, x2 Futhi njalo. Amanani abo langempela.

Ngokwesimo esisodwa nesezinhlobonhlobo eziningi ezingekho emgqeni, isilinganiso sokuhlanganiswa sibalwa:

-1 ≤ m ≤ 1;

0 ≤ m ≤ 1

Kukholelwa ukuthi ubuhlobo obuphakathi kwezinto ezikhiqizayo kanye nezici ezenzekayo ezifakwe emodeli buthaka uma inani le-tightness isilinganiso lokuxhumana (m) lingaphakathi kuka-0-0.3; Uma u-0,3-0,7 - ubukhulu bokuxhumana buphakathi; Ngaphezulu kuka 0.7-1 - ukuxhuma kunamandla.

Njengoba u-coefficient coefficient (ohlangene) p, i-coefficient yokuhlanganisa (multiple) P, i-correlation ratio m ibonakala, ngakho-ke ama coefficients of ukubaluleka kwawo abalwa kuwo (anqunywe kusuka ematafuleni). Uma lezi coefficients zikhulu kunani le-tabular, khona-ke ama coefficients tightness of uxhumano yizimbangela ezibalulekile. Uma ama-coefficients of materiality of the tightness of the uxhumano angaphansi kwamanani amisiwe noma uma coefficient ukuxhuma ngokwawo kungaphansi 0.7, ke imodeli akubandakanyi zonke izici ezithinta kakhulu umphumela.

I-coefficient yokuzimisela ibonisa ngokucacile ukuthi iziphi amaphesenti ezifakiwe kumodeli zinquma ukwakheka komphumela.

D = P2 * 100%

D = p2 * 100%

D = m2 * 100%

Uma i-coefficient yokuzikhethela ingaphezu kuka-50, khona-ke imodeli ichaza ngokufanele inqubo ephenywayo, uma ingaphansi kwama-50, kufanele sibuyele esiteji sokuqala sokwakhiwa futhi sibukeze ukukhethwa kwezinkomba zesici sokufakwa kumodeli.

Umklamo we-Fisher noma umgomo we-Fisher uveza ukusebenza komklamo wonke. Uma inani elibalwa le-coefficient lidlula inani elibekiwe, imodeli eyakhiwe ifanele ukuhlaziywa, kanye nezinkomba zokuhlela, izibalo zekusasa. Cishe inani letafula = 1.5. Uma inani elilinganisiwe lingaphansi kwenani elibekiwe, kuyadingeka ukwakha imodeli kuqala, kufaka phakathi nezici ezithinta kakhulu umphumela. Ngaphandle kokusebenza komklamo wonke, umthamo ngamunye we-coefficient uthonya izinto ezibonakalayo. Uma inani elibalwe laleli coefficient selidlulile inani elibekwe eceleni, i-coefficient yokuguqula iyoba yinto ephawulekayo, uma ingaphansi, khona-ke isici esilinganiselwa ngalo lo coefficient sisuswe, isamba siqala kuqala, kodwa ngaphandle kwalesi sici.