[Limdep Nlogit List] Error 1097 with random parameter logit model

[Yang Li]

Dr. Greene,

Thank you very much for your reply and your explanation!

If that is the situation, is there any good way to identify the significant
random parameters correctly among such a large variable set, when we cannot
have all parameters to be examined at the same time?

Thanks again for your help and your time!

Yang Li

On Sun, Jun 21, 2020 at 1:45 PM William Greene via Limdep <
limdep at mailman.sydney.edu.au> wrote:

> Yang Li. The internal limit is 25 RPs.  But, that is very optimistic.  The
> approach might work with 5 or 10 RPs, but beyond that, there must be
> identification issues.
> /B. Greene
>
> On Thu, Jun 18, 2020 at 2:13 PM Yang Li via Limdep <
> limdep at mailman.sydney.edu.au> wrote:
>
> > Dear all,
> >
> > I am new to NLOGIT and trying to fit my data with a random parameter
> logit
> > model.
> >
> > I've used a MNL to filter the significant variables for the next step of
> > running the ML model and everything went well with 59 variables at the
> end.
> >
> > But when I added the FCN command with all 59 parameters in it, this
> message
> > "Error 1097: Unusable LHS variable in choice model setup" appeared and
> the
> > program told me that all my observations flagged as bad.
> >
> > My code is:
> > NLOGIT;
> > LHS = DECISION;
> > CHOICES = KA, B, PO, NONE;
> > MODEL:
> > U(KA) = KO*ONE + k1*v1 + k3*V3 + k4*V4 + k6*V6 + k7*V7 +
> > k23*V23 + k24*V24 + k26*V26 + k27*V27 + k28*V28 + k30*V30 +
> > k31*V31 + k32*V32 + k35*V35 + k38*V38 +
> > k39*V39 + k40*V40 +
> > k46*V46 + k47*V47 + k50*V50 + k52*V52 +
> > k56*V56 + k65*V65/
> >
> > U(B) = BO*ONE + b1*v1 + b3*V3 + b4*V4 + b6*V6 + b7*V7 +
> > b23*V23 + b24*V24 + b27*V27 + b28*V28 +
> > b31*V31 + b32*V32 + b38*V38 +
> > b39*V39 + b40*V40 +
> > b46*V46 + b47*V47 +
> > b55*V55 +
> > b63*V63 + b65*V65/
> >
> > U(PO) = PO*ONE + p1*v1 + p3*V3 + p5*V5 + p7*V7 +
> > p10*V10 + p15*V15 +
> > p23*V23 + p27*V27 +
> > p37*V37 +
> > p39*V39 + p40*V40 + p41*V41 +
> > p46*V46 + p47*V47 + p50*V50 +
> > p55*V55 + p67*V67/
> >
> > U(NONE) = n13*V13;
> >
> >     OUTPUT = 1;
> >     MAXIT = 200;
> >     RPL;
> >     FCN = K1(N), K3(N), k4(N), K6(N), K7(N), K23(N), k24(N), K26(N),
> > k27(N),
> >     K28(N), k30(N), k31(N), K32(N), k35(N), K38(N), K39(N), K40(N),
> K46(N),
> >     K47(N), k50(N), k52(N), k56(N), k65(N), B1(N), B3(N), b4(N), b6(N),
> >     B7(N), b23(N), b24(N), b27(N), B28(N), b31(N), B32(N), B38(N),
> B39(N),
> >     B40(N), B47(N), b55(N), b63(N), b65(N), P1(N), P3(N), p5(N), P7(N),
> >     P10(N), P15(N), p23(N), p27(N), P37(N), p39(N), P40(N), p41(N),
> P46(N),
> >     p47(N), p50(N), p55(N), p67(N),n13(N);
> >     PTS = 25;
> >     HALTON$
> >
> > All my variables are 0-1 coded and choice invariant. The error will be
> gone
> > when I reduce the number of parameters that are put into the FCN command.
> > And the capability of the NLOGIT with ML model is described with up to
> 100
> > random parameters. Since I am able to fit my data with the ML model in
> SAS,
> > I am really confused with this error.  Any help would be appreciated and
> > thanks in advance!
> >
> > Yang Li
> > _______________________________________________
> > Limdep site list
> > Limdep at mailman.sydney.edu.au
> > http://limdep.itls.usyd.edu.au
> >
> >
>
> --
> William Greene
> Department of Economics, emeritus
> Stern School of Business, New York University
> 44 West 4 St.
> New York, NY, 10012
> URL: https://protect-au.mimecast.com/s/iUz1CwV1vMfYl3WATVxwri?domain=people.stern.nyu.edu
> Email: wgreene at stern.nyu.edu
> Ph. +1.646.596.3296
> Editor in Chief: Journal of Productivity Analysis
> Editor in Chief: Foundations and Trends in Econometrics
> Associate Editor: Economics Letters
> Associate Editor: Journal of Business and Economic Statistics
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au
>
>