Consistent behavior for fit classification when the response isn't a factor

[EmilHvitfeldt]

For whatever reason you might accidentally specify your response as something other then a factor. This gives rather informative error messages.

library(tidymodels)
#> ── Attaching packages ────────────────────────────────────────────────────────────────────── tidymodels 0.0.2 ──
#> ✔ broom     0.5.1     ✔ purrr     0.2.5
#> ✔ dials     0.0.2     ✔ recipes   0.1.4
#> ✔ dplyr     0.7.8     ✔ rsample   0.0.3
#> ✔ ggplot2   3.1.0     ✔ tibble    1.4.2
#> ✔ infer     0.4.0     ✔ yardstick 0.0.2
#> ✔ parsnip   0.0.1
#> ── Conflicts ───────────────────────────────────────────────────────────────────────── tidymodels_conflicts() ──
#> ✖ purrr::discard() masks scales::discard()
#> ✖ rsample::fill()  masks tidyr::fill()
#> ✖ dplyr::filter()  masks stats::filter()
#> ✖ dplyr::lag()     masks stats::lag()
#> ✖ recipes::step()  masks stats::step()

### factor

model_logistic <- logistic_reg(mode = "classification") %>%
  set_engine("glm") %>%
  fit(Species ~ ., data = iris)
#> Warning: glm.fit: algorithm did not converge
#> Warning: glm.fit: fitted probabilities numerically 0 or 1 occurred

predict_class(model_logistic, iris)
#>   [1] setosa     setosa     setosa     setosa     setosa     setosa    
#>   [7] setosa     setosa     setosa     setosa     setosa     setosa    
#>  [13] setosa     setosa     setosa     setosa     setosa     setosa    
#>  [19] setosa     setosa     setosa     setosa     setosa     setosa    
#>  [25] setosa     setosa     setosa     setosa     setosa     setosa    
#>  [31] setosa     setosa     setosa     setosa     setosa     setosa    
#>  [37] setosa     setosa     setosa     setosa     setosa     setosa    
#>  [43] setosa     setosa     setosa     setosa     setosa     setosa    
#>  [49] setosa     setosa     versicolor versicolor versicolor versicolor
#>  [55] versicolor versicolor versicolor versicolor versicolor versicolor
#>  [61] versicolor versicolor versicolor versicolor versicolor versicolor
#>  [67] versicolor versicolor versicolor versicolor versicolor versicolor
#>  [73] versicolor versicolor versicolor versicolor versicolor versicolor
#>  [79] versicolor versicolor versicolor versicolor versicolor versicolor
#>  [85] versicolor versicolor versicolor versicolor versicolor versicolor
#>  [91] versicolor versicolor versicolor versicolor versicolor versicolor
#>  [97] versicolor versicolor versicolor versicolor versicolor versicolor
#> [103] versicolor versicolor versicolor versicolor versicolor versicolor
#> [109] versicolor versicolor versicolor versicolor versicolor versicolor
#> [115] versicolor versicolor versicolor versicolor versicolor versicolor
#> [121] versicolor versicolor versicolor versicolor versicolor versicolor
#> [127] versicolor versicolor versicolor versicolor versicolor versicolor
#> [133] versicolor versicolor versicolor versicolor versicolor versicolor
#> [139] versicolor versicolor versicolor versicolor versicolor versicolor
#> [145] versicolor versicolor versicolor versicolor versicolor versicolor
#> Levels: setosa versicolor virginica

### logical

model_logistic <- logistic_reg(mode = "classification") %>%
  set_engine("glm") %>%
  fit(Species ~ ., data = iris %>% mutate(Species = Species == "setosa"))
#> Warning: glm.fit: algorithm did not converge

#> Warning: glm.fit: fitted probabilities numerically 0 or 1 occurred

predict_class(model_logistic, iris)
#> Warning in rep(yes, length.out = length(ans)): 'x' is NULL so the result
#> will be NULL
#> Error in ans[test & ok] <- rep(yes, length.out = length(ans))[test & ok]: replacement has length zero

### numeric

model_logistic <- logistic_reg(mode = "classification") %>%
  set_engine("glm") %>%
  fit(Species ~ ., data = iris %>% mutate(Species = as.numeric(Species)))
#> Error in eval(family$initialize): y values must be 0 <= y <= 1

### character

model_logistic <- logistic_reg(mode = "classification") %>%
  set_engine("glm") %>%
  fit(Species ~ ., data = iris %>% mutate(Species = as.character(Species)))
#> Error in eval(family$initialize): y values must be 0 <= y <= 1

^{Created on 2018-12-13 by the reprex package (v0.2.1)}

When response is logical we manage to fit, but predict_class fails. And when the response is numeric and character it fails to fit in both.

I propose we check inside fit.model_spec that the response is of type factor when doing classification.

[EmilHvitfeldt]

I don't mind trying to fix it if you deem it necessary.

[topepo]

(sorry @EmilHvitfeldt I didn't see your comment to volunteer until I started)

[EmilHvitfeldt]

No problem, I'm just happy it is being handled!

[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.