bind_rows: Expecting a string vector: [type=NULL; required=STRSXP].

[MatthieuStigler]

I am being stuck with dplyr 0.7.4 trying to use bind_rows() (initially: unnest()) on a list of data-frames obtained with broom::augment(), I get error message:
Error in bind_rows_(x, .id) :
Expecting a string vector: [type=NULL; required=STRSXP].

I don't see what is not working, it seems that this is due to a few offending columns... Am I doing something wrong or is there an issue down there?

library(tidyverse)
library(broom)

reg1 <- lm(y~ ., data=freeny)
reg2 <- lm(y~ . -price.index, data=freeny)
regs_df <- data_frame(name=c("reg1", "reg2"),
                      data=list(reg1, reg2)) %>%
  mutate(aug=map(data, augment))

## doesn't work?
regs_df$aug %>%
  bind_rows()

# seem to be linked to specific variables?
vars_issue <- c(2, 9, 12, 13)

regs_df$aug %>%
  map(~select(., -vars_issue)) %>%
  bind_rows()

## what is specific about these variables?
regs_df[1,]$aug[[1]] %>%
  select(vars_issue)

Thanks!!

[batpigandme]

Could you please submit this as a self-contained reprex (short for minimal reproducible example)?

If you've never heard of a reprex before, you might want to start by reading the tidyverse.org help page.

[mmstigler]

Oh, sorry about that! Here is the reprex version of the code:

library(tidyverse)
#> ── Attaching packages ────────────────────────────────── tidyverse 1.2.1 ──
#> ✔ ggplot2 2.2.1     ✔ purrr   0.2.4
#> ✔ tibble  1.4.2     ✔ dplyr   0.7.4
#> ✔ tidyr   0.7.2     ✔ stringr 1.2.0
#> ✔ readr   1.1.1     ✔ forcats 0.2.0
#> ── Conflicts ───────────────────────────────────── tidyverse_conflicts() ──
#> ✖ dplyr::filter() masks stats::filter()
#> ✖ dplyr::lag()    masks stats::lag()
library(broom)

reg1 <- lm(y~ ., data=freeny)
reg2 <- lm(y~ . -price.index, data=freeny)
regs_df <- data_frame(name=c("reg1", "reg2"),
                      data=list(reg1, reg2)) %>%
  mutate(aug=map(data, augment))

## doesn't work?
regs_df$aug %>%
  bind_rows()
#> Error in bind_rows_(x, .id): Expecting a string vector: [type=NULL; required=STRSXP].

# seem to be linked to specific variables?
vars_issue <- c(2, 9, 12, 13)

regs_df$aug %>%
  map(~select(., -vars_issue)) %>%
  bind_rows() %>%
  head
#>   .rownames lag.quarterly.revenue price.index income.level
#> 1   1962.25               8.79636     4.70997      5.82110
#> 2    1962.5               8.79236     4.70217      5.82558
#> 3   1962.75               8.79137     4.68944      5.83112
#> 4      1963               8.81486     4.68558      5.84046
#> 5   1963.25               8.81301     4.64019      5.85036
#> 6    1963.5               8.90751     4.62553      5.86464
#>   market.potential  .fitted     .se.fit      .hat     .sigma
#> 1          12.9699 8.789170 0.007561687 0.2636055 0.01493540
#> 2          12.9733 8.802520 0.006755339 0.2103834 0.01478897
#> 3          12.9774 8.821706 0.007021352 0.2272786 0.01488781
#> 4          12.9806 8.838953 0.006553851 0.1980205 0.01407319
#> 5          12.9831 8.883883 0.009562633 0.4215722 0.01393696
#> 6          12.9854 8.920665 0.006373890 0.1872950 0.01462400

## what is specific about these variables?
regs_df[1,]$aug[[1]] %>%
  select(vars_issue) %>%
  head
#>         y       .resid     .cooksd .std.resid
#> 1 8.79236  0.003189835 0.004560545  0.2523896
#> 2 8.79137 -0.011149901 0.038678336 -0.8519642
#> 3 8.81486 -0.006845771 0.016447669 -0.5287732
#> 4 8.81301 -0.025942587 0.191054486 -1.9669357
#> 5 8.90751  0.023627347 0.648560736  2.1093513
#> 6 8.93673  0.016065357 0.067482031  1.2099918

[romainfrancois]

The issue is these columns are flagged as objects:

> data <- regs_df$aug %>%
+   map(~select(., vars_issue))
> 
> data %>% map( ~map_lgl(., is.object) )
[[1]]
         y     .resid    .cooksd .std.resid 
      TRUE       TRUE       TRUE       TRUE 

[[2]]
         y     .resid    .cooksd .std.resid 
      TRUE       TRUE       TRUE       TRUE 

and not the others.

> regs_df$aug %>%
+   map(~select(., -vars_issue)) %>% map( ~map_lgl(., is.object) )
[[1]]
            .rownames lag.quarterly.revenue           price.index          income.level      market.potential               .fitted               .se.fit                  .hat                .sigma 
                FALSE                 FALSE                 FALSE                 FALSE                 FALSE                 FALSE                 FALSE                 FALSE                 FALSE 

[[2]]
            .rownames lag.quarterly.revenue           price.index          income.level      market.potential               .fitted               .se.fit                  .hat                .sigma 
                FALSE                 FALSE                 FALSE                 FALSE                 FALSE                 FALSE                 FALSE                 FALSE                 FALSE

fix on the way.

[MichaelChirico]

Seems like this will be fixed as a bug in R.

http://r.789695.n4.nabble.com/model-frame-strips-class-as-promised-but-fails-to-strip-OBJECT-in-C-td4748687.html

[lock]

This old issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with reprex) and link to this issue. https://reprex.tidyverse.org/