Hello Members,
I am a new Stata user and now trying to write a paper about prediction model. But we meet a problem now. As we had missing data in this analysis, we used multiple imputation to produce an overall estimate of regression coefficient (with syntax 'mi estimate : logistic' )and model performance measurements including C-statistic (with syntax 'mi xeq 0: roctab ', the probability is also an overall probability of all imputation calculated with code 'mi predict xb_mi' and 'mi xeq: generate phat = invlogit(xb_mi)') and calibration in the large and calibration slope (with syntax 'mi xeq 0:pmcalplot ') . But when we try to draw nomogram with syntax nomolog (with syntax 'mi xeq 0:nomolog ', maybe it is not the correct syntax?), the STATA indicated that nomolog must be executed after logit or logistic. But there were 10 imputation, and the model is the overall estimated model.
Would you please teach me how to draw a nomogram in the multiple imputated database?
Many thanks and best wishes!
Jing Pan
I am a new Stata user and now trying to write a paper about prediction model. But we meet a problem now. As we had missing data in this analysis, we used multiple imputation to produce an overall estimate of regression coefficient (with syntax 'mi estimate : logistic' )and model performance measurements including C-statistic (with syntax 'mi xeq 0: roctab ', the probability is also an overall probability of all imputation calculated with code 'mi predict xb_mi' and 'mi xeq: generate phat = invlogit(xb_mi)') and calibration in the large and calibration slope (with syntax 'mi xeq 0:pmcalplot ') . But when we try to draw nomogram with syntax nomolog (with syntax 'mi xeq 0:nomolog ', maybe it is not the correct syntax?), the STATA indicated that nomolog must be executed after logit or logistic. But there were 10 imputation, and the model is the overall estimated model.
Would you please teach me how to draw a nomogram in the multiple imputated database?
Many thanks and best wishes!
Jing Pan
Comment