[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

From |
Liqun An <an_liqun@yahoo.com> |

To |
statalist@hsphsun2.harvard.edu |

Subject |
st: Question on maximum likelihood estimation |

Date |
Sun, 22 Jun 2003 01:00:03 -0700 (PDT) |

Dear List, I am wondering if anyone could give me suggestions on the following codes that I wrote. It seems that "ml search" or "ml check" could never find a feasible value to start. The error message is "could not find feasible values". I have checked the codes many times but still couldn't figure out what goes wrong. Any suggestion would be greatly appreciated. Frank ************************************************ /* This is to construct our own likelihood function and maximize it*/ /* the data used is "C:\DATA\dataact.dta"*/ clear set more off cap log close set mem 300m cd c:\psid log using "C:\psid\data16",replace t capture program drop jtest program define jtest version 6 args todo b lnf tempvar theta1 theta2 tempvar z1 z2 z3 z4 z5 z12 a e D tempvar lt tempname s_H alpha lsh lalp mleval `theta1' = `b', eq(1) mleval `theta2' = `b', eq(2) /*hazard rate*/ mleval `s_H' = `b', eq(3) scalar mleval `alpha' = `b', eq(4) scalar quietly{ scalar `s_H' = `s_H' scalar `alpha' = `alpha' scalar `lsh' = ln(`s_H') scalar `lalp' = ln(`alpha') gen double `z1' = ln(normd(($ML_y1-`theta1')/(`s_H')))-`lsh' /*normal*/ gen double `z5' = -(`theta2')*(($ML_y2)^(`alpha')) gen double `z2' = ln(`theta2') + `lalp'+ (`alpha'-1)*ln($ML_y2)+`z5' /*weibull*/ gen double `z12' = `z1'+`z2' by newid: gen `lt' = cond(_n==_N, _N,.) gen `D' = (`lt'==1) if `lt' !=. by newid: gen double `z3' = cond(_n==(_N-1), sum(`z12'),.) by newid: gen double `z4' = `z3'[_n-1] if (`lt'!=.&`lt'!=1) gen double `e' = (1-$ML_y3)*(`z12')+($ML_y3)*(`z5') by newid: gen double `a'=`z4'+`e' if (`lt'!=.&`lt'!=1) mlsum `lnf' = `D'*`e'+(1-`D')*`a' if `lt'!=. } end use "C:\DATA\dataact.dta", clear /*gen lhv = ln(hv)*/ sort newid t save "C:\DATA\dataact.dta", replace use "C:\DATA\dataact.dta", clear ml model d0 jtest (lhv dur c = fsize)(age)/s_H /alpha __________________________________ Do you Yahoo!? SBC Yahoo! DSL - Now only $29.95 per month! http://sbc.yahoo.com * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

- Prev by Date:
**st: sample selection bias** - Next by Date:
**st: Analysis of proportion as the dependant variable** - Previous by thread:
**st: sample selection bias** - Next by thread:
**st: Analysis of proportion as the dependant variable** - Index(es):

© Copyright 1996–2014 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |