Stata da Logit va Probit regression

Logit va Probit — natija 0 yoki 1 bo'lganda ishlatiladi (ish bor/yo'q, kasalmi/yo'q).

Logit regression

logit ish_bor ta_lim yosh tajriba jinsi

* Natija: log-odds koeffitsientlar

Odds Ratio

logit ish_bor ta_lim yosh, or
* yoki
logistic ish_bor ta_lim yosh

Probit regression

probit ish_bor ta_lim yosh tajriba jinsi

Marginal effektlar (muhim!)

Logit/Probit koeffitsientlar to'g'ridan-to'g'ri izohlanmaydi, marginal effektlar kerak:

* O'rtachada marginal effektlar (MEM)
margins, dydx(*)

* O'rtacha marginal effektlar (AME)
margins, dydx(*) atmeans

* Muayyan qiymatda
margins, dydx(ta_lim) at(yosh=30 jinsi=1)

Klassifikatsiya va model sifati

logit ish_bor ta_lim yosh
estat classification
estat gof          * Hosmer-Lemeshow testi
lroc               * ROC egri chizig'i

Tobit model (cheklangan natija)

* Daromad 0 dan past bo'lolmasa
tobit daromad ta_lim yosh, ll(0)
margins, dydx(*)