①please tell me the case of following ERROR
ERROR: A problem occured during model estimation. Diagnostic information is as follows:
Matrix not positive definite
OCCURRED IN: cholesky_in_place in MOD logit_model
Logistic regression failed at iteration 1
-Hessian not positive definite.
OCCURRED IN: run_logit_model_threads in MOD logit_model
Could not fit logit model. The problem occurred in:
Group: 1.
OCCURRED IN: mstep_beta in MOD estimation
M-step for BETA failed at iteration 1.
Could not fit logit model. We suggest
using the BETA PRIOR statement.
OCCURRED IN: run_em_modelfit in MOD estimation
②for example, when I want to incorporate all of the following variables,proc lca program is right?
House number Customer number employer number sex height(cm) weight(kg) product year period
112 1 LO2113 1 161 58 A0001 2018 3
112 2 LO2223 2 140 60 A0002 2017 4
113 3 LO2113 1 131 73 A0001 2013 2
113 4 LO2113 2 170 121 A0003 2014 6
221 5 LO2004 2 121 48 A0002 2015 7
222 6 LO2113 2 135 50 A0002 2017 8
223 7 LO2113 2 140 . A0002 2016 1
PROC LCA DATA=out OUTPARAM=param1 OUTPOST=post1 OUTEST=est1 OUTSTDERR=stderr1;
NCLASS 5;
id employer number;
item sex;
CATEGORIES 2;
COVARIATES height weight year period;
clusters house number ;
NSTARTS 50;
RHO PRIOR=1;
SEED 1000;
RUN;
- id：Is multile variables is available?
2. when I want to standarize, would I better to use not covariates statement,but weight statement?
3. what statement do I use to incorporate product variables?
