(single) record view

[elbersb]

This may have come up before, but I couldn't find anything in the tibble package or here.

It might be desirable to implement a printing mode that "transposes" the tibble. This came up when demonstrating bloom::glance on a presentation slide, where the output looks like this:

glance(mod)

#> # A tibble: 1 x 12 
#> r.squared adj.r.squared sigma statistic p.value df 
#>     <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> 
#>       1 0.652 0.652 7.62 3192. 0 1 
#> # ... with 6 more variables: logLik <dbl>, AIC <dbl>, 
#> # BIC <dbl>, deviance <dbl>, df.residual <int>, 
#> # nobs <int>

(the white space got messed up a bit when copying from the PDF)

In this case, all columns are numeric, so the matrix transpose doesn't look so bad:

> t(glance(mod))
                       [,1]
r.squared         0.7168580
adj.r.squared     0.7165251
sigma             6.8773735
statistic      2153.2928344
p.value           0.0000000
df                2.0000000
logLik        -5702.0854721
AIC           11412.1709442
BIC           11433.9338790
deviance      80454.3501172
df.residual    1701.0000000
nobs           1704.0000000

However, tibble could of course support this more fully, including a column for the column type and applying the correct formatting. A similar feature exists in PostgreSQL: https://stackoverflow.com/questions/9604723/alternate-output-format-for-psql. I always found this nice to have.

One could even go so far as treating a single row like a "record", and defaulting to printing it this way. But that might be confusing.

[krlmlr]

Thanks. Does tibble::glimpse() help with that?

[elbersb]

Sorry, should've mentioned glimpse. This is in fact pretty similar to glimpse, but in its current form, not quite the same. I think there are three issues: glimpse prints all columns by default (so if your data frame has a 1,000 columns, glimpse will print 1,000 lines); glimpse doesn't apply the same useful formatting that print does; and the biggest one is that if you have more than one row, the columns are not aligned:

d <- data.frame(x = c(1, 100), y = c(100, 1))
dplyr::glimpse(d)
#> Rows: 2
#> Columns: 2
#> $ x <dbl> 1, 100
#> $ y <dbl> 100, 1

^{Created on 2021-05-21 by the reprex package (v2.0.0)}

(I realize that this is probably a pretty niche issue, so feel free to close if you think this is out of scope.)

[krlmlr]

Thanks. Should we add an align = FALSE and max = Inf argument to glimpse() ?

I don't understand "the same useful formatting that print() does", can you please rephrase?

Other than that, I think most if not all of the tools to build custom views of the data are exported by pillar, please let me know if anything's missing.

[elbersb]

Yes, align = FALSE and max = Inf would go a long way!

Regarding the second point, glimpse doesn't apply the same formatting that print does. For instance, in the following example the labelled column is displayed differently, and the rounding is not applied for the numeric vector. The negative numbers also show up with red font color in the print call, but not in glimpse (of course, this is not visible in the reprex).

library("tibble")
df <- tibble(
  s = haven::labelled(c("M", "M", "F"), c(Male = "M", Female = "F")),
  x = rnorm(3)
)
df
#> # A tibble: 3 x 2
#>   s                x
#>   <chr+lbl>    <dbl>
#> 1 M [Male]    0.0849
#> 2 M [Male]    0.842 
#> 3 F [Female] -0.481
glimpse(df)
#> Rows: 3
#> Columns: 2
#> $ s <chr+lbl> "M", "M", "F"
#> $ x <dbl> 0.08492588, 0.84195136, -0.48107764

^{Created on 2021-05-25 by the reprex package (v2.0.0)}

[moodymudskipper]

The glimpse() function is used to show conveniently the head of the data, similar to str() on data frames.

We wish to implement align and max arguments that would permit printing the full transposed data in a transposed.

The 'terse' package does something similar: https://github.com/coolbutuseless/terse