R - Model with a lot of dummy variables

linear-regression, model, r

Solution

R will create dummy variables for you automatically, here is a basic example:

> mycars <- mtcars
> mycars$cyl <- factor(mycars$cyl)
> fit <- lm( mpg ~ wt+cyl, data=mycars, x=TRUE )
> summary(fit)

Call:
lm(formula = mpg ~ wt + cyl, data = mycars, x = TRUE)

Residuals:
    Min      1Q  Median      3Q     Max 
-4.5890 -1.2357 -0.5159  1.3845  5.7915 

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  33.9908     1.8878  18.006  < 2e-16 ***
wt           -3.2056     0.7539  -4.252 0.000213 ***
cyl6         -4.2556     1.3861  -3.070 0.004718 ** 
cyl8         -6.0709     1.6523  -3.674 0.000999 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 2.557 on 28 degrees of freedom
Multiple R-squared: 0.8374,     Adjusted R-squared:  0.82 
F-statistic: 48.08 on 3 and 28 DF,  p-value: 3.594e-11 

> head(fit$x)
                  (Intercept)    wt cyl6 cyl8
Mazda RX4                   1 2.620    1    0
Mazda RX4 Wag               1 2.875    1    0
Datsun 710                  1 2.320    0    0
Hornet 4 Drive              1 3.215    1    0
Hornet Sportabout           1 3.440    0    1
Valiant                     1 3.460    1    0
>

The `x=TRUE` in the call to `lm` tells it to return the x matrix actually used, which includes the dummy variables. If you don't want to look at the created dummy variables then you can leave that out. See `?contrasts` for more detail if you want to set how the dummy variables are created.

Problem

If I have a column in a data set that has multiple variables how would I go about creating these dummy variables. Example: Lets say that I have a column named color it has: Red, Green, Yellow, Blue, Pink, and Grey as options for the color of a car. What is the best way to turn these variables into factors. without creating a bunch of dummy variables by hand? Edit: So I did what Greg recommended and this is what I have. I was wondering about the NA output though and was unsure why it is there. ``` > data$Trim<-factor(data$Trim) > data$Model<-factor(data$Model) > data$Type<-factor(data$Type) > data=cbind(Price,Mileage,Buick,Cadillac,Chevrolet,Pontiac,SAAB,Saturn,Model,Trim,Type,Cylinder,Liter,Doors,Cruise,Sound,Leather) > fit <- lm( Price ~ Mileage+Buick+Cadillac+Chevrolet+Pontiac+SAAB+Saturn+Model+Trim+Type+Cylinder+Liter+Doors+Cruise+Sound+Leather, x=TRUE ) > summary(fit) ``` Then I get a message "Coefficients: (21 not defined because of singularities)" and for some of the variables the output is NA.

Original source