Izindlela Zokulinganisa Ezibalweni
Izibalo ziyisayensi yokuqoqa, ukuhlaziya, nokuhumusha idatha, futhi enye yezingxenye zayo ezibalulekile ukulinganisa. Ukulinganisa kuzibalo kubhekisela enkambisweni yokunquma inani elilinganiselwe lepharamitha yenani labantu ngokusekelwe kulwazi olutholwe kusampula. Izindlela zokulinganisa zingahlukaniswa ngezinhlobo ezimbili eziyinhloko: ukulinganisa amaphuzu kanye nokulinganisa izikhawu. Kulesi sihloko, sizoxoxa ngezindlela ezahlukene zokulinganisa ezivame ukusetshenziswa kuzibalo.
Ukuqonda Okuyisisekelo Kokulinganisa
Ngaphambi kokuthi singene ezindleleni zokulinganisa, kubalulekile ukuqonda amagama ayisisekelo:
– Amapharamitha: Izici zezinombolo zenani labantu. Isibonelo, isilinganiso sabantu (µ), ukuhlukahluka kwabantu (σ²).
– Izibalo: Izici zezinombolo zesampula. Isibonelo, isilinganiso sesampula (x̄), ukuhlukahluka kwesampula (s²).
Umgomo oyinhloko wokulinganisa ukwenza iziphetho mayelana namapharamitha esibalo sabantu ngokusekelwe kudatha yesampula. Kunezinhlobo ezimbili eziyinhloko zokulinganisa kuzibalo:
1. Isilinganiso Sephuzu: Inikeza inani elilodwa kuphela njengesilinganiso sepharamitha yenani labantu.
2. Ukulinganisa Kwesikhawu: Kuhlinzeka ngamanani ahlukahlukene njengesilinganiso sepharamitha yenani labantu, kufaka phakathi izinga elithile lokuzethemba.
Indlela Yokulinganisa Amaphuzu
Ukulinganisa amaphuzu inqubo yokuhlinzeka ngenombolo eyodwa eyisilinganiso esihle kakhulu sepharamitha yenani labantu. Ezinye izilinganiso zamaphuzu ezisetshenziswa kakhulu yilezi:
1. Isilinganiso (Isilinganiso) sesampula
Indlela elula nevame kakhulu yokulinganisa isilinganiso sabantu iwukusebenzisa isilinganiso sesampula, esibalwa kanje:
\[ \bar{x} = \frac{1}{n} \sum_{i=1}^{n} x_i \]
lapho \( x_i \) kungukubonwa ngakunye kusampula kanye \( n \) kungusayizi wesampula.
2. Isampula Ephakathi
I-median yesampula iyinani eliphakathi ledatha yesampula ehlungiwe. Iyisikali esinamandla ngoba ayithintwa yizinto ezingaphandle.
3. Isilinganiso Sesampula
Ukuze kulinganiswe isilinganiso sabantu, kusetshenziswa isilinganiso sesampula, esibalwa kanje:
\[ \hat{p} = \frac{x}{n} \]
lapho \( x \) kuyinani lempumelelo kusampula kanye \( n \) kuyisayizi yesampula.
Indlela Yokulinganisa Isikhawu
Izilinganiso zesikhashana zinikeza ububanzi bamanani, okulindeleke ukuthi zimboze ipharamitha yabantu ngezinga elithile lokuzethemba (isb., 95%). Izilinganiso zesikhashana zivame ukuvezwa ngesimo se-Confidence Interval (CI).
1. Isikhawu Sokuqiniseka Sesilinganiso Sabantu
Uma idatha yesampula ivela ekusabalalisweni okuvamile noma i-\( n \) inkulu ngokwanele (kusebenza i-CLT), isikhawu sokuzethemba sesilinganiso sabantu \( \mu \) singu:
\[ \bar{x} \pm z_{\alpha/2} \frac{\sigma}{\sqrt{n}} \]
Kuphi:
– \( \bar{x} \) isilinganiso sesampula
– \( z_{\alpha/2} \) yinani lika-z lokusabalalisa okujwayelekile okuhambisana nezinga lokuzethemba (isb., 1.96 ngo-95%)
– \( \sigma \) ukuphambuka kwesilinganiso sabantu. Uma \( \sigma \) kungaziwa, \( s \) (isampula yokuphambuka kwesilinganiso) iyasetshenziswa.
– \( n \) usayizi wesampula.
2. Isikhawu Sokuqiniseka Ngenani Labantu
Ukulinganisa isilinganiso sabantu \( p \):
\[ \hat{p} \pm z_{\alpha/2} \sqrt{\frac{\hat{p}(1-\hat{p})}{n}} \]
lapho \( \hat{p} \) kuyisilinganiso sesampula kanye neminye imingcele njengoba kuchaziwe ngaphambilini.
Ezinye Izindlela Zokulinganisa
1. Indlela Yokungenzeka Okukhulu (ML)
Indlela ye-Maximum Likelihood iyindlela esetshenziswa ukuthola isilinganiseli esingcono kakhulu sepharamitha yabantu \( \theta \) ngokwandisa umsebenzi we-likelihood \( L(\theta) \). Umsebenzi we-likelihood ungamathuba okuthola idatha ebonwe inikezwe ipharamitha \( \theta \):
\[ L(\theta|x) = \prod_{i=1}^{n} f(x_i|\theta) \]
lapho \( f(x_i|\theta) \) kungumsebenzi wobuningi be-likelihood (PDF) yedatha. Isikali esikhulisa \( L(\theta) \) sibizwa ngokuthi isikali se-likelihood esiphezulu (MLE).
2. Indlela Yokulinganisa yeBayesian
Indlela yeBayesian iphatha amapharamitha njengezinto eziguquguqukayo ezingahleliwe futhi isebenzisa ukusatshalaliswa kwamathuba ukulinganisa amapharamitha. Ngokusho kweTheorem kaBayes:
\[ P(\theta|x) = \frac{P(x|\theta) P(\theta)}{P(x)} \]
lapho \( P(\theta|x) \) kungukusatshalaliswa kwangemuva, \( P(x|\theta) \) kungukuthi kungenzeka, \( P(\theta) \) kungukuthi kwenzeka ngaphambili, kanye \( P(x) \) kungukuthi kungenzeka. Izilinganiso ze-Bayesian zincike kakhulu kulokho okusetshenzisiwe.
Ukuhlolwa Kokulinganisa
Ukuze sihlole isilinganiso samaphuzu, kumele sihlole izakhiwo zaso:
– Ukulunga/Ukubandlulula: Umqageli \( \hat{\theta} \) kuthiwa akabandlululi uma \( E[\hat{\theta}] = \theta \).
– Ukusebenza Kahle: Isikali esisebenzayo sinokuhlukahluka okuncane kakhulu phakathi kwazo zonke izikali ezingachemile.
– Ukuvumelana: Kuthiwa isilinganiso siyavumelana uma \( \hat{\theta} \) sisondela \( \theta \) njengoba usayizi wesampula \( n \) ukhuphuka.
Izibonelo Zokusebenza
1. Isilinganiso Semali Engenayo
Ocwaningweni lwezomnotho, ukulinganisa imali engenayo evamile yabantu kuvame ukwenziwa. Abacwaningi bathatha isampula yabantu bese bebala isilinganiso sesampula njengesilinganiselo samaphuzu, bese benikeza isikhawu sokuzethemba ukuze babonise ukungaqiniseki kwalesi silinganiso.
2. Ukuqagela Isilinganiso Sabavoti
Kuhlolovo lokhetho, umcwaningi angase afune ukulinganisa iphesenti labavoti abasekela umuntu othile omele ukhetho. Isilinganiso sesampula \( \hat{p} \) sabaphenduli abasekela lowo muntu sisetshenziswa njengesikali samaphuzu. Isikhawu sokuzethemba singanikezwa ukukhombisa umkhawulo wephutha.
Isiphetho
Izindlela zokulinganisa zidlala indima ebalulekile kwizibalo ngoba zivumela abacwaningi ukuthi benze iziphetho mayelana nezibalo zabantu ngokusekelwe kudatha yesampula. Izindlela zokulinganisa amaphuzu nezikhawu zinikeza amathuluzi anamandla alokhu, ngamasu anjengokungenzeka okukhulu kanye nokulinganisa kwe-Bayesian okujula ekubunzimeni kwedatha. Ukusebenzisa izilinganiso ezifanele, ezisebenzayo, nezihambisanayo kuqinisekisa imiphumela yokuhlaziywa kwedatha ethembekile nenembile, okwenza kube lula ukwenza izinqumo ezingcono emikhakheni efana nezomnotho, isayensi yezenhlalo, impilo, nokuningi.
