[R] update fit (removing insignificant variables)

arun smartpink111 at yahoo.com
Wed Nov 21 22:48:51 CET 2012


HI,

I am not sure about the ?update() method.
You could try this:

set.seed(232)
mat1<-matrix(sample(1:100,80,replace=TRUE),ncol=8) #with 8 columns
dat1<-data.frame(mat1)
 names(dat1)[1]<-"Y"
 fit<-lm(Y~.,data=dat1)
res<-coef(summary(fit))

 res
#                 Estimate Std. Error     t value   Pr(>|t|)
#(Intercept) 143.763081344 21.9902865  6.53757200 0.02260698
#X2           -0.009030461  0.1103611 -0.08182650 0.94223654
#X3           -1.844809745  0.2655513 -6.94709326 0.02009766
#X4            1.436647351  0.2540660  5.65462216 0.02988003
#X5           -0.005406684  0.2906822 -0.01859998 0.98684896
#X6           -0.050101169  0.1486149 -0.33712071 0.76811703
#X7            0.563284987  0.1204945  4.67477901 0.04284010
#X8           -1.901918462  0.2697166 -7.05154460 0.01952387

Wtd<-row.names(res[res[,"Pr(>|t|)"]<0.05,])
 Wtd1<-Wtd[!grepl("\\(.*\\)",Wtd)]
fit1<-lm(as.formula(paste("Y~",paste(Wtd1,collapse="+"))),data=dat1)
 coef(summary(fit1))
#               Estimate  Std. Error   t value     Pr(>|t|)
#(Intercept) 146.8921451 10.87053736  13.51287 3.975711e-05
#X3           -1.9002023  0.10667909 -17.81232 1.023637e-05
#X4            1.4421062  0.09790384  14.72982 2.606788e-05
#X7            0.5726324  0.05396720  10.61075 1.285606e-04
#X8           -1.9560052  0.14204298 -13.77052 3.625118e-05

A.K.
________________________________
From: farnoosh sheikhi <farnoosh_81 at yahoo.com>
To: arun <smartpink111 at yahoo.com> 
Sent: Wednesday, November 21, 2012 2:47 PM
Subject: update fit (removing insignificant variables)


Hi there,

I have a question for removing insignificant predictors from fit.
The output looks like below but with 170 predictors.
Lab_1                                                              0.037928 *  
Lab_2                                                              0.030423 *  
Lab_4                                                              0.616122    
Lab_7                                                              0.101278 
Lab_8                                                              2.44e-06 ***
Lab_12                                                             0.003181 ** 
Lab_14                                                             0.096809 

I need to update my model with significant ones.
I know if I want to remove one, I use the following code, but I don't know how to do that for 70 insignificant predictors.
fit2<-update(fit, .~. -Lab_7)



Thanks a lot.



Best,Farnoosh Sheikhi 



More information about the R-help mailing list