Statalist The Stata Listserver


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

Re: st: hit rate after logit model


From   Jeph Herrin <jherrin@flyingbuttress.net>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: hit rate after logit model
Date   Fri, 02 Jun 2006 12:29:39 -0400

For starters, I think you want to use -forvalues- instead of -foreach-.

Because you do -tab- after -sum-, r(mean) is lost. Try:

  qui sum hit
  local mu=r(mean)
  tab `e(depvar)' pred, freq nocol
  di in green "the mean hit rate is `mu'"

if you want to store the mean, or just reverse the order of the
-sum- and the -tab-.

Hope this helps.

Jeph


Dirk Nachbar wrote:
All

here is a program (the last for this week I promise) which is supposed
to run after various choice models. It should calculate the hit rate of
correct predictions. Two things create problems: the mean hit rate is
not shown and the -foreach- loop doesn't seem to work. Maybe someone
would like to take a closer look.

clear

set obs 100

gen x=invnorm(uniform())

gen e=invnorm(uniform())

gen y=round(2*exp(1+x+e)/(1+exp(1+x+e)),1)

ologit y x

/*

Program to run after most Discrete Choice Models

it reports the hit rate of correct predictions

*/

capture program drop postdcm

program define postdcm

version 8.2

syntax [if] [in] [, threshold(real 0.5)]

if `threshold'<0 | `threshold'>1 {

    di as error "threshold has to be between 0 and 1"

    exit 198

    }

if "`e(cmd)'"=="" {

    di as error "no DCM estimations found"

    exit 198

    }

if "`e(cmd)'"=="logit" | "`e(cmd)'"=="probit" | "`e(cmd)'"=="xtlogit" |
"`e(cmd)'"=="xtprobit" {

    tempvar pred pred2

    qui predict pred

    gen pred2=(pred>=`threshold')

    gen hit=(`e(depvar)'==pred2)

    sum hit

    tab `e(depvar)' pred2, freq nocol

    di in green "the mean hit rate is `r(mean)'"

    }

if "`e(cmd)'"=="ologit" | "`e(cmd)'"=="oprobit" | "`e(cmd)'"=="mlogit" |
"`e(cmd)'"=="mprobit" {

    qui sum `e(depvar)'

    gen pred=`r(min)'

    foreach i=`r(min)'/`r(max)' {

        tempvar pred`i'

        qui predict pred`i', outcome(`i')

        if `i'==`r(min)' {

            gen pred=`i'

            local max=pred`i'

            }

        else if pred`i'>`max' {

            replace pred=`i'

            local max=pred`i'

            }

        }

    gen hit=(`e(depvar)'==pred)

    qui sum hit

    tab `e(depvar)' pred, freq nocol

    di in green "the mean hit rate is `r(mean)'"

    }

end

postdcm

Dirk Nachbar

Assistant Economist

Pensim2

Department for Work and Pensions

Level 4, The Adelphi

1-11 John Adam St

WC2N 6HT London

020 796 28531
**********************************************************************
This document is strictly confidential and is intended only for use by
the addressee. If you are not the intended recipient, any disclosure,
copying, distribution or other action taken in reliance of the
information contained in this e-mail is strictly prohibited. Any views
expressed by the sender of this message are not necessarily those of the
Department for Work and Pensions. If you have received this transmission
in error, please use the reply function to tell us and then permanently
delete what you have received. Please note: Incoming and outgoing e-mail
messages are routinely monitored for compliance with our policy on the
use of electronic communications.
**********************************************************************
The original of this email was scanned for viruses by Government Secure
Intranet (GSi) virus scanning service supplied exclusively by Cable &
Wireless in partnership with MessageLabs.
On leaving the GSI this email was certified virus free.
The MessageLabs Anti Virus Service is the first managed service to
achieve the CSIA Claims Tested Mark (CCTM Certificate Number
2006/04/0007), the UK Government quality mark initiative for information
security products and services. For more information about this please
visit www.cctmark.gov.uk

*
*   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/



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