Confidence lokuphumla. Yintoni na kwaye zingasetyenziswa njani na?

Confidence lokuphumla, kwanawezayo kuthi, evela entsimini sezibalo. Le uluhlu ethile, nelisebenza ukuqikelela i parameter ezingaziwayo ngezinga eliphezulu lokuthembeka. Indlela elula ukucacisa oku ngokwenza umzekelo.

Masithi ufuna ukuhlola naliphi na ixabiso random, umz, ixesha lokuphendula umncedisi kwisicelo client. Lonke ixesha echwetheza umsebenzisi idilesi ethile, umncedisi aphendule kuyo ngezantya ezahlukeneyo. Ngoko, ixesha lokuphendula yokuhlola random. Ngoko ke, lo lokuphumla ukuzithemba ukufumanisa nemida yalo parameter, yaye ke ngoko kuya kuba lula ukuphikisana ukuba kukho amathuba okuba 95% izinga abasabela ngayo kumncedisi kuba kuluhlu ebalwe kuthi.

Okanye ufuna ukwazi abantu abaninzi banolwazi trade mark yenkampani. Xa lokuphumla ukuzithemba ibalwa, ngoko kuya kuba kunokwenzeka, umzekelo, ukuba bathi ukuba 95% amathuba umlinganiselo abathengi bayazi kule lophawu, kukuba kuluhlu ukususela-27% ukuya ku-34%.

Ekubeni lo gama izalana ngokusondeleyo ixabiso efana inqanaba ukuzithemba. Oku kungenzeka ukuba ukhetho enqwenelekayo ifakiwe lokuphumla ngokuzithemba. Ukusuka eli xabiso kuxhomekeke big luya kuba kuluhlu lwethu ezinqwenelekayo. Okuya uba mkhulu umsantsa ixabiso elizifumeneyo, sincinane, ubude, kunye vice versa. Ngokuqhelekileyo oko imiselwe ku-90%, 95% okanye 99%. Ixabiso 95% yaziwa kakhulu.

Icandelo Active kuchaphazela ukuchithachitheka imigqaliselo kunye nobukhulu isampuli le. njengengcaciso lusekelwe kwintekelelo yokuba lesiphumo uchaphazelekayo ngokuxhomekeke umthetho lonikezelo oluqhelekileyo. Le ngxelo eyaziwa ngokuba uMthetho GAUSS kaThixo. Ngokutsho kwakhe, oku kubizwa ngokuba unikezelo eqhelekileyo variable rhoqo ejikelezayo linokuchazwa yi loxinano namathuba. Ukuba Kwingcingane lonikezelo oluqhelekileyo yaba ephosakeleyo, ngoko uqikelelo kungaba engalunganga.

Okokuqala, makhe kujongana ukubala indlela lokuphumla ukuzithemba kuba kulindelekileyo. Kukho iimeko ezimbini kunokwenzeka. Usasazeko (degree of Wazibhangisa le random bume) ukuze kwaziwa okanye hayi. Ukuba iyaziwa, interval sithembele ibalwa ngokusebenzisa le ndlela ilandelayo:

HSR - t * σ / (SQRT (n)) <= α <= t HSR + * σ / (SQRT (n)), apho

α - uphawu,

t - parameter le Laplace distribution table,

SQRT (n) - ingcambu ka'B lilonke umthamo isampula ,

σ - ingcambu ye IP.

Ukuba Ukunxaxha ayaziwa, oko kungaba ibalwe, ukuba siyazi onke amaxabiso kwento efunekayo. Ukuze wenze oku, sebenzisa ifomula ilandelayo:

σ2 = h2sr - (HSR) 2, apho

h2sr - ixabiso avareji lwezikweri kwento wafunda,

(HSR) 2 - square kuthetha ixabiso le isici.

Ifomula ngalo kule meko ubalwa ukuzithemba lokuphumla yahluke kancinane:

HSR - t * s / (SQRT (n)) <= α <= HSR + t * s / (SQRT (n)), apho

XCP - isampuli ukuthini,

α - uphawu,

t - parameter efumaneka yi ukwabiwa Student itafile t = t (ɣ; n-1);

SQRT (n) - ingcambu ka'B ubungakanani isampulu lo,

s - ingcambu ye IP.

Xiya xikombiso lexi. Ukucinga ukuba iziphumo ngokwemilinganiso 7 wayezimisele ixabiso umndilili test inkalo, leyo ilingana ku-30 kunye mahluko isampuli ilingana 36. Kufuneka kungafunyanwa amathuba ka-99% ukuzithemba lokuphumla leyo iqulathe ixabiso yokwenene ye parameter kulinganiswa.

Okokuqala thina ezichaza ukuba yintoni na le t: t = t (0,99; 7-1) = 3.71. Ukusebenzisa indlela ngasentla, sifumana:

HSR - t * s / (SQRT (n)) <= α <= HSR + t * s / (SQRT (n))

30 - 3.71 * 36 / (SQRT (7)) <= α <= 30 + 3,71 * 36 / (SQRT (7))

21,587 <= α <= 38,413

Isithuba yokuqiniseka mahluko ibalwa njengoko kunjalo eyaziwayo kuthetha, kwaye xa kungekho data kwi Ithemba zemathematika, kunye nexabiso eyaziwayo ukulinganisela mahluko cala ngongoma kuphela. Asiyi kubanika apha ifomula ukubala wayo, ekubeni zintsonkothe kakhulu kwaye, ukuba oyifunayo, ukuba kusoloko afumaneka kwi-network.

Siphawula kuphela ukuba lokuphumla ukuzithemba wagqiba ngokusebenzisa inkqubo okanye isevisi yenethwekhi Excel, ekuthiwa yi encedisayo.