[Limdep Nlogit List] Clarification on "Error 806: Line search does not improve fn. Exit iterations. Status=3"

[William Greene]

Saidur. You have too many constant terms. The separate branch models
each must be equivalent to an MNL, so you have to drop the constants
on taxi and tran. We'll see if you have to drop the one on walk also.
I don't think so, since the probability is 1.00 for walk given branch.
There is no constant term for the branch, so BW may be identified.
/Bill Greene

----- Original Message -----
From: "Saidur Rahman Milan" <srmilan at gmail.com>
To: limdep at limdep.itls.usyd.edu.au
Sent: Tuesday, July 1, 2008 7:38:53 AM (GMT-0500) America/New_York
Subject: [Limdep Nlogit List] Clarification on "Error 806: Line search does not improve fn. Exit iterations. Status=3"

*
Dear All,

I am facing  problems in running a program for two-level NL Model. I need
your cordial help to solve such problems. While running the program, errors
occurs like as :


Line search does not improve fn. Exit iterations. Status=3

Error 806: Line search does not improve fn. Exit iterations. Status=3

Function= .82017889876D+04, at entry, .78639135462D+04 at exit

Error 1025: Failed to fit model. See earlier diagnostic.
Sometimes, I get results with a report like "Maximum iteration reached. Exit
iterations with status=1" after 101 iteration completion. Please help me
solve the problem. I herewith mention all the codes and results below:


Thanks,
Saidur Rahman
Hiroshima University, Japan
========================


NLOGIT ; Lhs = mode

; Choices= tran,rick,taxi,cng,mc,walk,pc

; Tree = PV(pc,mc),CAP1(taxi,cng),cap2(tran,rick), CAP3(walk)

; Model: U(pc) = BTIME*TIME+BCOST*COST/

U(mc) =BM+BTIME*TIME+BCOST*COST/

U(tran) =BT+BTIME*TIME+BCOST*COST/

U(rick) =BR+BTIME*TIME+BCOST*COST/

U(Walk) =BW+BTIME*TIME+BCOST*COST/

U(taxi) =BX+BTIME*TIME+BCOST*COST/

U(cng) =BG+BTIME*TIME+BCOST*COST $
*

Normal exit from iterations. Exit status=0.

+---------------------------------------------+

| Discrete choice (multinomial logit) model |

| Maximum Likelihood Estimates |

| Model estimated: Jul 01, 2008 at 08:27:51PM.|

| Dependent variable Choice |

| Weighting variable None |

| Number of observations 6826 |

| Iterations completed 7 |

| Log likelihood function -8201.789 |

| R2=1-LogL/LogL* Log-L fncn R-sqrd RsqAdj |

| No coefficients -13869.8751 .40866 .40849 |

| Constants only -8800.8822 .06807 .06780 |

| Chi-squared[ 2] = 1198.18635 |

| Prob [ chi squared > value ] = .00000 |

| Response data are given as ind. choice. |

| Number of obs.= 6826, skipped 0 bad obs. |

+---------------------------------------------+

+---------+--------------+----------------+--------+---------+

|Variable | Coefficient | Standard Error |b/St.Er.|P[|Z|>z] |

+---------+--------------+----------------+--------+---------+

BTIME -.07334947 .00267528 -27.417 .0000

BCOST -.01658187 .00123668 -13.408 .0000

BM -2.95382058 .15921637 -18.552 .0000

BT 1.17038186 .09474623 12.353 .0000

BR 2.16408933 .08057977 26.856 .0000

BW 1.60071129 .11237772 14.244 .0000

BX 3.76596762 .24099093 15.627 .0000

BG -.74883756 .08786534 -8.523 .0000

Line search does not improve fn. Exit iterations. Status=3

Error 806: Line search does not improve fn. Exit iterations. Status=3

Function= .82017889876D+04, at entry, .78639135462D+04 at exit

Error 1025: Failed to fit model. See earlier diagnostic.


-- 
Saidur Rahman Milan
Room #: 301(Transportation Engineering Lab)
International Development and Cooperation (IDEC)
Hiroshima University
1-5-1 Kagamiyama, Higashi-Hiroshima
739-8529 JAPAN
Phone: 0081-90-6430-7214 (Mobile), 0081-82-422-6555 (Home)
Web URL: http://srmilan.tripod.com/, http://home.hiroshima-u.ac.jp/hitel/
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au


-- 
Professor William Greene
Department of Economics
Stern School of Business
New York University
44 West 4th St., Rm. 7-78
New York, NY   10012
http://www.stern.nyu.edu/~wgreene
212.998.0876