CODEWITHSUNDEEP
rdata.tablelapplysummary
Ian Wang
Ian Wang

Reputation: 157

Using lapply in data.table to extract element from a list column

I merged a list of lm summaries with a data.table. Next, I extract a coefficient from the summaries and generate a new column containing the number. My question is, why do I have to use lapply to complete such work?
For example, the following code worked.

new.DT <- old.DT[, result := lapply(X = results.of.lm, FUN = summary)] %>%
  .[, beta := lapply(X = result, function(x) x$coefficients[2,1])]

While the following code failed.

new.DT <- old.DT[, result := lapply(X = results.of.lm, FUN = summary)] %>% 
.[, beta := result$coefficients[2,1])]

However, for some functions I apply to columns in data.table (e.g. paste, substr or as.numeric), lapply was not necessary. I couldn't figure out what cause the difference. Thank you!

Upvotes: 0

Views: 559

Answers (1)

langtang
langtang

Reputation: 24722

Presuming that old.DT looks something like this:

   grp results.of.lm
1:   1      <lm[12]>
2:   2      <lm[12]>
3:   3      <lm[12]>

then, after creating the result column using old.DT[, result:=lapply(results.of.lm,summary)], we note that result is a list of lists

   grp results.of.lm           result
1:   1      <lm[12]> <summary.lm[11]>
2:   2      <lm[12]> <summary.lm[11]>
3:   3      <lm[12]> <summary.lm[11]>

Note that old.DT$result$coefficients is NULL, and thus beta:=result$coefficients[2,1] will NOT return the desired result. Instead, your use of lapply does return the coefficients:

old.DT[, beta:=lapply(result, function(x) x$coefficients[2,1])][]

   grp results.of.lm           result       beta
1:   1      <lm[12]> <summary.lm[11]> -0.2819342
2:   2      <lm[12]> <summary.lm[11]>  0.1645671
3:   3      <lm[12]> <summary.lm[11]>  0.2215897

Upvotes: 1

Related Questions