tibble name repair with logistic regression

[bcjaeger]

cc @DavisVaughan

library(tidymodels)
#> -- Attaching packages ------------------------------------------------------------------------- tidymodels 0.0.2 --
#> v broom     0.5.1     v purrr     0.2.5
#> v dials     0.0.2     v recipes   0.1.4
#> v dplyr     0.7.8     v rsample   0.0.4
#> v ggplot2   3.1.0     v tibble    2.0.1
#> v infer     0.4.0     v yardstick 0.0.2
#> v parsnip   0.0.1
#> -- Conflicts ---------------------------------------------------------------------------- tidymodels_conflicts() --
#> x purrr::discard() masks scales::discard()
#> x dplyr::filter()  masks stats::filter()
#> x dplyr::lag()     masks stats::lag()
#> x recipes::step()  masks stats::step()

analysis=data.frame(target=sample(c(0,1),100,replace=T),
                    x1=rnorm(100),
                    x2=rnorm(100))

head(analysis)
#>   target         x1         x2
#> 1      1 -0.5611995  1.4630014
#> 2      1  1.7830997 -1.6015657
#> 3      1  1.2500967 -1.5411327
#> 4      1 -0.4197018 -0.4608742
#> 5      0 -0.4842685  0.1226822
#> 6      1  0.7493446 -1.2447351

dsplit <- initial_split(analysis, strata = 'target')

trn=training(dsplit)
tst=testing(dsplit)

lrm_fit <- logistic_reg() %>% 
  set_engine('glm') %>%
  fit(target~x1+x2,data=trn) 

lrm_fit %>%
  predict(tst, type='prob')
#> Error: Columns 1, 2 must be named.
#> Use .name_repair to specify repair.

^{Created on 2019-01-16 by the reprex package (v0.2.1)}

[topepo]

I think that the issue here is the nature of the outcome data. We want to coerce encourage people to use better data structures (factors here instead of numeric 0/1).

There is an issue open that will check for factor outcome data in classification models for parsnip.

[bcjaeger]

That makes sense. I'm guessing this has already been implemented, but I would say that an informative error or warning message would make this a closed issue for me. For example, instead the error message above, it would be great to see an error message like Error: the outcome column, (paste name here), should be a factor for classification problems. If this is already in the code, great! If not, feel free to take or leave my suggestion.

[topepo]

Fixed way back ...

library(tidymodels)
#> Registered S3 method overwritten by 'xts':
#>   method     from
#>   as.zoo.xts zoo
#> ── Attaching packages ────────────────────────────────────────────────────────────────────────────────────────────── tidymodels 0.0.3 ──
#> ✔ broom     0.5.2          ✔ purrr     0.3.3     
#> ✔ dials     0.0.3.9001     ✔ recipes   0.1.7.9001
#> ✔ dplyr     0.8.3          ✔ rsample   0.0.5     
#> ✔ ggplot2   3.2.1          ✔ tibble    2.1.3     
#> ✔ infer     0.5.0          ✔ yardstick 0.0.4     
#> ✔ parsnip   0.0.3.9001
#> ── Conflicts ───────────────────────────────────────────────────────────────────────────────────────────────── tidymodels_conflicts() ──
#> ✖ purrr::discard()  masks scales::discard()
#> ✖ dplyr::filter()   masks stats::filter()
#> ✖ dplyr::lag()      masks stats::lag()
#> ✖ ggplot2::margin() masks dials::margin()
#> ✖ dials::offset()   masks stats::offset()
#> ✖ recipes::step()   masks stats::step()

analysis=data.frame(target=factor(sample(c(0,1),100,replace=T)),
                    x1=rnorm(100),
                    x2=rnorm(100))

head(analysis)
#>   target          x1          x2
#> 1      1  0.12830796 -0.99972044
#> 2      0 -0.03901003  0.82826500
#> 3      0  0.27802064 -0.08822105
#> 4      1 -1.59099101  1.50020975
#> 5      0 -0.21978763  1.08039572
#> 6      0 -0.06999034 -0.92425982

dsplit <- initial_split(analysis, strata = 'target')

trn=training(dsplit)
tst=testing(dsplit)

lrm_fit <- logistic_reg() %>% 
  set_engine('glm') %>%
  fit(target~x1+x2,data=trn) 

lrm_fit %>%
  predict(tst, type='prob')
#> # A tibble: 24 x 2
#>    .pred_0 .pred_1
#>      <dbl>   <dbl>
#>  1   0.532   0.468
#>  2   0.529   0.471
#>  3   0.567   0.433
#>  4   0.350   0.650
#>  5   0.589   0.411
#>  6   0.544   0.456
#>  7   0.486   0.514
#>  8   0.486   0.514
#>  9   0.468   0.532
#> 10   0.536   0.464
#> # … with 14 more rows

^{Created on 2019-11-01 by the reprex package (v0.3.0)}

[bcjaeger]

Thank you!! =]

Sorry I did not close it when the fix was applied, my mistake.

[github-actions]

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