Linear Model Loop in R

Hello,

I have a list of linear model outputs in R:

Models <- c('FGm', 'ECm', 'BWm', 'Mm', 'BDm', 'Cm', 'WCm', 'FCm',
                'DPC1m','DPC2m','DPC3m','DPC4m','DPC5m','DPC6m','DPC7m',
                'DPC8m','DPC9m','DPC10m','SPC1m','SPC2m','SPC3m','SPC4m',
                'SPC5m','SPC6m','SPC7m','SPC8m','SPC9m','SPC10m', 'Ipom', 
                'Mertm', 'Seasm')

I am trying to generate a column with residual and predicted values in the dataframe:

for (i in Models){
  res <- paste('Data$', i,".res", sep="")
  pred <- paste('Data$', i,".pred", sep="")
  res <- residuals(i)
  pred <- predict(i)
}

I keep getting the following error, can someone help with this?

Error: $ operator is invalid for atomic vectors

• 1,795 views • link

updated 3.2 years ago by Ram

• written 5.7 years ago by selplat21 •

Try:

res <- Data[ paste0(i,".res") ]
pred <- Data[ paste0(i,".pred") ]

• 0 views • link

written 5.7 years ago by zx8754

That is helpful, thank you! The problem here is that the following concatenated list is in character format, but I want each item in this list to be treated as an lm type object. So, not sure how to just get these read in as lm objects.

Models <- c('FGm', 'ECm', 'BWm', 'Mm', 'BDm', 'Cm', 'WCm', 'FCm',
                'DPC1m','DPC2m','DPC3m','DPC4m','DPC5m','DPC6m','DPC7m',
                'DPC8m','DPC9m','DPC10m','SPC1m','SPC2m','SPC3m','SPC4m',
                'SPC5m','SPC6m','SPC7m','SPC8m','SPC9m','SPC10m', 'Ipom', 
                'Mertm', 'Seasm')

for (i in Models){
  Data[ paste0(i,".res") ] <- residuals(i)
}

• 0 views • link

written 5.7 years ago by selplat21 •

Beyond that, are you aware that the first res and pred are being overwritten in the 3th and 4th line of the for loop body without seemingly being used before?

• 0 views • link

written 5.7 years ago by ATpoint

So here was my rationale here:

Models <- list(FGm, ECm, BWm, Mm, BDm, Cm, WCm, FCm, 
                DPC1m,DPC2m,DPC3m,DPC4m,DPC5m,DPC6m,DPC7m,
                DPC8m,DPC9m,DPC10m,SPC1m,SPC2m,SPC3m,SPC4m,
                SPC5m,SPC6m,SPC7m,SPC8m,SPC9m,SPC10m, Ipom, 
                Mertm, Seasm) #list of lm type objects

Sorry, I realized the overwrite. The following should illustrate what I want to do:

for (i in Models){
  paste0('Data$', names(Models)[i],".res", sep="")  <- residuals(i) 
}

The problem here is that I'm unable to access the names of the list items from above. This term "names(Models)[i]" won't work unfortunately.

• 0 views • link

written 5.7 years ago by selplat21 •

1 answer

Suppose you have a list of lm models.

Models <- list(lm(mpg~cyl, mtcars), lm(hp~drat, mtcars))

When you have a list, you generally use lapply to iterate over it and apply a function.

Data <- lapply(Models, function(x) data.frame(res=residuals(x), pred=predict(x)))

I wasn't completely sure from your description what you wanted as an output, but I output a list of data.frames containing the residuals and predicted values for each model.

> lapply(Data, head)
[[1]]
                         res     pred
Mazda RX4          0.3701643 20.62984
Mazda RX4 Wag      0.3701643 20.62984
Datsun 710        -3.5814159 26.38142
Hornet 4 Drive     0.7701643 20.62984
Hornet Sportabout  3.8217446 14.87826
Valiant           -2.5298357 20.62984

[[2]]
                         res     pred
Mazda RX4         -19.226118 129.2261
Mazda RX4 Wag     -19.226118 129.2261
Datsun 710        -39.103380 132.1034
Hornet 4 Drive    -66.413209 176.4132
Hornet Sportabout   2.614957 172.3850
Valiant           -89.827684 194.8277

• 0 views • link

written 5.7 years ago by rpolicastro

You are an amazing human, thank you so much. This is perfect.

• 0 views • link

written 5.7 years ago by selplat21 •