Ukungavumelani - imodeli unxulumano uphawu

model ukuhlanganisa (CM) - inkqubo ubalo enika yokwamkela quadratic zezibalo, apho isalathisi onemveliso kubalulwe kuxhomekeke zalathisi omnye okanye ngaphezulu.

yx = ao + a1h1

apho: y - neziphawuli zokusebenza, kuxhomekeke x factor;

x - factor umqondiso;

A1 - Ipharamitha KM, ebonisa utshintsho kangakanani kwi Isalathisi onemveliso xa ukutshintsha factor x ngabanye, ukuba zonke ezinye izizathu ezichaphazela y azitshintshanga;

AO CM parameter obonisa isiphumo zonke ezinye izinto kwisalathiso eyiyo y, ngaphandle factor bume x

Xa ukhetha izalathisi olusebenzayo kunye neemodeli factor kufuneka ithathele ingqalelo into yokuba nezibonisi zomsebenzi ezikwi of kudaleka limi phezu kwinqanaba eliphezulu ngaphezu factor yokusebenza.

Features model ukuhlanganisa

Emva yokubala ukungavumelani imodeli ukuhlanganisa iparameters ibalwe.

p - simple ukungavumelani, -1 ≤ r ≤ 1, ibonisa amandla kunye nolwalathiso Isalathisi kwi nefuthe factor amanqaku. Ukusondela 1, ngokuqina ulwalamano, ukusondela 0, ubuhlobo ithambile. Ukuba inani ukuhlanganisa umntu lowo unesi sifo, ngoko ukuxhumeka tye, ukuba awunayo - kubuyela ebutsheni.

Ukungavumelani Ifomula: pxy = (x-x * 1 / y) / * eu eh

eh = hh2- (x) 2; eu = y2 (y) 2

Ukuba linear CM multifactorial, ukuba ifomu:

yx = ao + a1h1 a2x2 + + ... + anx

ke yabalelwa ukungavumelani ezininzi.

0 ≤ P ≤ 1, kwaye ibonisa amandla nefuthe zonke izibonisi factor amanqaku athathwe kunye.

P = 1- ((yi-yi) 2 / (yi -usr) 2)

Apho: uh - Isalathisi ezinemveliso - ixabiso kubalwa;

yi - ixabiso elililo;

ixabiso usr- kanye, avareji.

Uqikelelo yi ixabiso afunyenwe kufakwe imodeli ukuhlanganisa endaweni x1, x2 njl izimiso zazo actual.

Njengezibonelo univariate kwaye multivariate nonlinear ukuhlanganisa ratio ibalwa:

-1 ≤ m ≤ 1;

0 ≤ m ≤ 1

Kukholelwa ukuba ubudlelwane phakathi olusebenzayo zibandakanywe imodeli kwezalathisi impembelelo buthathaka, ukuba yengxinano le lomlingani coupling (m) kuluhlu 0-0.3; ukuba 0.3-0.7 - ngokusondela ulwalamano - umlinganiselo; 0.7-1 ngasentla - isibopho.

Ekubeni ukungavumelani (umphunga) r, ngumlingani ukuhlanganisa (multiple) R, ukuhlanganisa ratio m - ixabiso lemizuzu, ebalwayo ukuba okuza ngokokubaluleka kwazo (kuchaziwe iitafile). Ukuba ezi okuza kuba ngaphezu ngokwexabiso itafile, ngokusondela okuza uqhagamshelwano oonobangela kubalulekile. Ukuba okuza kufuphi nokuhlangana essentiality yengxinano ezincinane ngaphezu amaxabiso itheyibhile okanye ukuba self coupling lomlingani ngaphantsi kwe 0.7, imodeli ayiquki yonke impembelelo iparameters ezichaphazela ngokuphawulekayo isiphumo.

'Ngumlingani ka ukuzimisela ubonisa ipesenteji zifakiwe kwi parameters imodeli ukuqinisekisa ukusekwa yesiphumo factor.

D = P2 * 100%

D = P2 * 100%

D = m 2 * 100%

Ukuba inani kukuzimisela mkhulu kwe-50, ngoko ke imodeli uchaza ngokwaneleyo inkqubo phantsi isifundo, ukuba ngaphantsi kwe-50, kuyimfuneko ukuba uye emva kwinqanaba lokuqala zokwakha, kwaye nokuhlaziya izalathi factor akhethwe ngenjongo yokufakwa kule modeli.

Fisher Fisher factor okanye kwingongoma olwahlula ukusebenza kwale modeli xa iyonke. Xa umlinganiselo ubalwe mkhulu ngaphezu kwetafile, imodeli eyakhelwe ilungele ukuba uhlalutyo kwakunye njengezalathisi yocwangciso zokusonjululwa elizayo. Lwaba ixabiso itafile = 1.5. Ukuba ixabiso ubalwe lingaphantsi kwe etafileni, kufuneka kuqala ukwakha imodeli, kuquka imiba ebalulekileyo eziphembelela isiphumo. Ukongeza ukusebenza le modeli ngokubanzi kuchaphazela kakhulu ingumlingani parameter nganye. Ukuba ixabiso ebalwe kule ratio kwadlulwa kwitheyibhile ubukhulu, ngumlingani parameter nganye kuyaphawuleka ukuba ngaphantsi, ngoko ke parameter factor, apho inani abaliweyo zisuswe izibalo isampuli uqale kwakhona, kodwa ngaphandle kule inkalo.