Multiple regressions on sets of different independent variables

Question

At the beginning: I am very 'fresh' user of R so excuse me all my mistakes and silly questions.

I would like to estimate 570 (30*19) equations (and get summary() of them). Each equation is in the form: y~x2+x3+x4, where x2, x3 and x4 are the list.

In comment #1 is my failed trial - it can show what i want to achieve. In comment #2 such equation is working but I need to manually change variable "jan" (to feb, mar, and so on) that should be taken from list "x2". This code return me 30 equations but I need to estimate each of them with one more variable - x2 from the list of 19's elements named "x2"

How to do it automatically ? I mean: mapply - applies the rlm model to each of the element of the lists so, e.g: element 1 from list "y" is corresponding with element 1 from "x3" and element 1 from "x4" (I took care about the order when I was preparing csv file) - this is correct but I would like to run such equations like below comment #2 with additional variable x2. This variable is not corresponding any other variables - it always should be in each of the 30 regressions of elements of list "y", just changing from element 1 of list "x2" to element 19 of "x2".

In the end I would like to have 570 (30*19) "summ" matrix and 570 "archtest"

I do hope this is not to muddled, if yes I will try to post it again in some time with clarification.

Thank you kindly for understanding and help.

data<-read.table("MAY EFFECT Rdata 18.09.2013-3 ind prod CCI.csv", head=T, sep=";", dec=".")
library(MASS)
library(FinTS)
index1<-data[,2:31]   # y  30  elements; rates, 30 countries
index2<-data[,32:50] # x2  19  elements; month dummies, months and other
index3<-data[,51:80] # x3  30  elements; IP
index4<-data[,81:110] # x4 30  elements; CCI

y<-as.list(index1)
x2<-as.list(index2)
x3<-as.list(index3)
x4<-as.list(index4)

#1. this is my trial I need to make "x2" variable that stands for month dummy be respectively equal jan, feb ... and so on till
# the end of list x2 (19 elements)
result<-mapply(function(x2) mapply(function(y,x3,x4,x2) summary(rlm(y~x2+x3+x4, maxit = 15600, data=data)), y,x3,x4),x2)

#2. this code below is returning 10x30 list matrix "summ" and 5x30 data frame "archtest"
summ<- mapply(function(y,x3,x4,jan) summary(rlm(y~jan+x3+x4, maxit = 15600, data=data)), y,x3,x4)
archtest<-mapply(function (y,x3,x4,jan,resi) {regr<-rlm(y~jan+x3+x4, maxit = 15600, data=data)
                                              ArchTest (resid(regr), lags=12, demean = FALSE)},y,x3,x4)

####

Answer 1

If I understand you correctly, there are 30 sets of (y, x3, and x4) and for each of these 19 sets of x2. If that's correct, then something like this should work:

set.seed(1)
data <- matrix(rnorm(50*110),nrow=50)  # completely random data...

get.summary <- function(i,j) {
  dta <- data.frame(y=data[,1+i],x2=data[,31+j], x3=data[,50+i],x4=data[,80+i])
  summary(lm(y~x2+x3+x4, data=dta))
}

smry.list <- do.call(cbind,lapply(1:19,function(j){lapply(1:30,get.summary,j)}))
smry.list[1,1]

So now smry.list contains all 570 summaries, arranged in 19 columns by 30 rows.

You can do something similar with mapply(...) but the result is a one-dimensional list:

smry.mapply<- mapply(get.summary,rep(1:30,each=19),1:19, SIMPLIFY=F)
smry.mapply[1]