CODEWITHSUNDEEP
r
overwhelmed
overwhelmed

Reputation: 225

How to reuse the call function in r?

Stepwise selection method (stepAIC in MASS library) provide the subset of variables in its last step. Instead of typing the final model I would like to use it through call function. I don't know how to do that! Following is an example: 

x = as.data.frame(matrix(cbind(rnorm(100), rnorm(100, 10), rnorm(100, 100), rpois(100, 4)), 100, 4))
fit = lm(V1 ~., data=x)
st = stepAIC(fit, direction = "both")
st$call
#  lm(formula = V1 ~ 1, data = x)

How to use st$call for further use? I tried the following with a hope that it will give me the regression results, but it's not working: 

fm = st$call
summary(call)

Any help is appreciated.

Upvotes: 1

Views: 166

Answers (2)

Dason
Dason

Reputation: 61933

The selected model is returned. You can just use st directly. The returned object has a little bit more information attached to it since it was the result of the stepwise procedure but the returned object can and should do everything that you'd want to do from fitting that model 'by hand'.

Upvotes: 0

loki
loki

Reputation: 10350

You can reuse a function call with eval().

data(iris)
model <- lm(Sepal.Length~., data = iris)
model
# Call:
#   lm(formula = Sepal.Length ~ ., data = iris)
# 
# Coefficients:
#   (Intercept)        Sepal.Width       Petal.Length        Petal.Width  Speciesversicolor   Speciesvirginica  
# 2.1713             0.4959             0.8292            -0.3152            -0.7236            -1.0235  


eval(model$call)
# Call:
#   lm(formula = Sepal.Length ~ ., data = iris)
# 
# Coefficients:
#   (Intercept)        Sepal.Width       Petal.Length        Petal.Width  Speciesversicolor   Speciesvirginica  
# 2.1713             0.4959             0.8292            -0.3152            -0.7236            -1.0235

As you can see the output of both calls are the same.

Upvotes: 2

Related Questions