Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Boosted regression


From   Megan Valentine <Megan.Valentine@newcastle.edu.au>
To   statalist@hsphsun2.harvard.edu
Subject   st: Boosted regression
Date   Tue, 11 Dec 2012 16:09:38 +1100

Dear Stata Users
I am trying to use boosted regression in STATA 11, following the example set by Shenyang Guo to conduct some Propensity Score Analysis.
I have included Guo's example code and output and my problem code below. The influence should add up to 100%, but mine is 0% and one variable has no influence recorded. the variables are a mixture of continuous, dichotomous and ordinal variables, and service (and kuse) are treatment (Y/N) variables.  
Any thoughts??
Thanks
Meg


Test Example

Code
boost kuse pcg_adc age97 mratio96 pcged97 black, distribution(logistic) maxiter(1000) trainfraction(0.8) pred(ps) int er(4)  shrink(.0005) influence

Output
influence
Distribution=logistic
predict=ps
Trainfraction=.8 Shrink=.0005 Bag=.5 maxiter=1000 Interaction=4
Fitting ... 
Assessing Influence ... 
Predicting ... 
bestiter= 1000
Test R2= .13907204
trainn= 802
Train R2= .18065613
Influence of each variable (Percent):
pcg_adc 4.75571 
age97 1.4344578 
mratio96 88.972091 
pcged97 2.9557829 
black 1.8819586 


My problem example

Code
boost service prem multbirth zbwz teachyear schinvolv homeck teachqual parentpar dremote2 male disab repeated daanga dacons heavydr dwphysoi dsep ageyears faredn,   distribution(logistic)  maxiter(1000) trainfraction(0.8) pred(ps) inter(4)  shrink(.0005)  influence

Output
influence
Distribution=logistic
predict=ps
Trainfraction=.8 Shrink=.0005 Bag=.5 maxiter=1000 Interaction=4
Fitting ... 
Assessing Influence ... 
Predicting ... 
bestiter= 1000
Test R2= .04586068
trainn= 1549
Train R2= .07024484
Influence of each variable (Percent):
prem 0 
multbirth 0 
zbwz . 
teachyear 0 
schinvolv 0 
homeck 0 
teachqual 0 
parentpar 0 
dremote2 0 
male 0 
disab 0 
repeated 0 
daanga 0 
dacons 0 
heavydr 0 
dwphysoi 0 
dsep 0 
ageyears 0 
faredn 0



Megan Valentine
Discipline of Statistics
School of Mathematical & Physical Sciences
Science Offices
University of Newcastle
megan.valentine@newcastle.edu.au


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


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