3.4.1 Uhlalutyo ukuqokelelwa kweenkcukacha ze data: Okungathi isampuli

PlanetsWeight ungenza ingenzeki iimposiso ngabom zibangelwa inkqubo isampulu.

Iisampuli Okungathi zezo apho bonke abantu babe eyaziwa, non-zero yindlela ukuqukwa, kwaye uyilo ilula linokuba ukukhiwa elula isampuli ejikelezayo apho umntu ngamnye amathuba ngokulinganayo ukubandakanywa. Xa abaphenduli zikhetwa ngokusebenzisa isampulu elula random ngokumbulala ofezekileyo (umzekelo, akukho imposiso iTshathi akukho non-impendulo), ukulinganisela kuthe tye kuba kwisampulu iya-ngokomyinge-kuba uguqulelo zancinane abemi.

Simple isampuli ejikelezayo ayifane ezisetyenziswa ke izenzo. Kunoko, abaphandi ngabom ukhethe abantu yenzekile yokungalingani ukufakwa ukwenzela ukunciphisa iindleko nokunyusa ukuchaneka. Xa abaphandi khetha abantu yenzekile ezahlukeneyo yokubandakanywa ngabom, koko kufuneka utshintsho ukukhulula kwezithintelo okubangelwa inkqubo isampulu. Ngamanye amazwi, indlela iimeko ukusuka isampuli kuxhomekeke indlela ekhethiweyo isampulu.

Ngokomzekelo, Current Population Survey (CPS) isetyenziswa ngurhulumente US ukuqikelela izinga lentswela. nyanga nganye abantu malunga-100,000 indlebe, nokuba ubuso-ngobuso okanye ngefowuni, kwaye iziphumo zisetyenziswa ukuvelisa uqikelelo izinga lentswela. Ngenxa yokuba urhulumente unqwenela ukuqikelela izinga ngqesho karhulumente ngamnye, alikwazi ukwenza isampuli elula random abadala kuba kungakhupha abaphenduli bambalwa kakhulu States kwabemi amancinane (umzekelo, Rhode Island) abaninzi kakhulu ukusuka States abemi abaninzi (umz , California). Endaweni yoko, iCPS iisampulu abantu kumazwe awahlukeneyo ngamazinga ahlukeneyo, inkqubo ebizwa isampuli nehlelwe kunye nokulindeleka yokungalingani ukhetho. Umzekelo, ukuba iCPS babefuna abaphenduli 2,000 karhulumente ngamnye, ngoko abadala Rhode Island kuba ngamaxesha 30 eziphezulu yindlela ukufakwa ngaphezu abadala California (Rhode Island: abaphenduli 2,000 ngeveki abadala 800.000 vs California: abaphenduli 2,000 ngeveki abadala 30,000,000). Njengoko siza kubona kamva, olu hlobo zokusampula kunye ezinokwenzeka olungalinganiyo ntoni na imithombo intanethi data kakhulu, kodwa ngokungafaniyo iCPS, indlela yeesampulu idla akwaziwa okanye elawulwa umphandi.

Ngenxa eyilwe isampulu, iCPS ayikho ngqo ummeli US; oku kuquka abantu abaninzi kakhulu ukusuka Rhode Island kwaye bambalwa kakhulu eCalifornia. Ngoko ke, kuya kuba bubulumko ukuqikelela izinga lentswela kweli lizwe izinga lentswela kwisampulu. Endaweni yokusetyenziswa kuthetha isampulu, kungcono ukuthatha kuthetha marks, apho amatye akhawunti ngenxa yokuba abantu Rhode Island ekwakunokulindeleka ukuba zibandakanywe kunabantu eCalifornia. Ngokomzekelo, umntu ngamnye ukususela California kuba upweighted- baza ndizibalile ngakumbi kuqikelelo-kwaye umntu ngamnye ukususela Rhode Island babeza downweighted-babeza abala ngaphantsi kulo uqikelelo. Enyanisweni, ukuba banikwa ilizwi ngakumbi ebantwini ukuba akulula ukuba bafunde.

Lo mzekelo yokudlala ubonisa ingongoma ebalulekileyo kodwa ngokuqhelekileyo abawuqondi: akuyomfuneko isampuli ukuba uguqulelo zancinane abemi ukuze avelise uqikelelo elungileyo. Ukuba ngokwaneleyo yaziwa ngendlela ndawonye i data, ngoko ukuze ulwazi ingasetyenziswa xa kusenziwa uqikelelo kwisampula. Le ndlela Ndikhe nje ochazwe-yaye ukuba ukuchaza ngokwezibalo kwiinkalo zobugcisa isihlomelo-iwela ngqo ukuba ngokwesakhelo lemizuzu isampuli yamandulo. Ngoku, ndiza kukubonisa ukuba ingcamango efanayo linokusetyenziswa iisampuli non-ezinokwenzeka.