CODEWITHSUNDEEP
r
BSM
BSM

Reputation: 11

Multivariate multiple linear regression with many dependent variables

How does one run multivariate linear regressions if there are a lot of dependent variables, say in my case 222 to be exact? I want to regress certain values I have for 222 different companies with few regressors.

I know i can do e.g.

y <- cbind(y1, y2, y3... yn)
fit <- lm(y ~ X1 + X2 + ... Xn)

But there has to be a clever way to cbind my columns other than writing by hand cbind(y1, y2, y3, ...y222) - right?

I have tried cbind(vol[, 2:223]) but placing that in y and given to lm() function only results in Error in model.frame.default(formula = y ~ RMF + SMB + HML, drop.unused.levels = TRUE) : invalid type (list) for variable 'y'

Not very experienced with R so I appreciate all the help I can get for my thesis! Please bear with me.

Upvotes: 0

Views: 2315

Answers (2)

G. Grothendieck
G. Grothendieck

Reputation: 269824

Below we use the built-in anscombe data frame as an example.

1) The key part is to use a matrix, not a data frame, for the left hand side of the formula. In the example below we define a matrix y of the dependent variables and then use that with lm:

y <- as.matrix(anscombe[5:8])
lm(y ~ x1 + x2 + x3 + x4, anscombe)

1a) or if there are many independent variables too:

lm(y ~ ., anscombe[1:4])

2) One could alternately use lm.fit. Note that it does not automatically add an intercept so we add one:

m <- as.matrix(anscombe)
lm.fit(cbind(Intercept = 1, m[, 1:4]), m[, 5:8])

lm.fit returns a list rather than an lm object but some methods such as coef and resid (but not summary) work with it anyways.

Upvotes: 1

YOLO
YOLO

Reputation: 21729

Assuming, all 222 vectors are of same length and your global environment has no other object except these 222 vectors, you can try:

Method 1:

library(purrr)

# get all 223 vectors in a list
vec_list <- as.list(.GlobalEnv))

# cbind the list elements
df <- map_df(vec_list, cbind)

Method 2: 

# this lists all the objects in your current environment
vec_list= ls()

# get data
df <- cbind(vec_list)

Upvotes: 0

Related Questions