model-ksvm.md
In tidypredict: Run Predictions Inside the Database

errors on unsupported models

Code
  tidypredict_fit(rbf)
Condition
  Error in `parse_model()`:
  ! Only linear kernlab SVM models are supported.
  i The model must use the "vanilladot" (linear) kernel.

Code
  tidypredict_fit(multi)
Condition
  Error in `parse_model()`:
  ! Only binary classification kernlab SVM models are supported.
  i This model has 3 classes.

Code
  tidypredict_fit(noprob)
Condition
  Error in `parse_model()`:
  ! Classification kernlab SVM models require a probability model.
  i Refit with `prob.model = TRUE`.

a factor level that is not a syntactic name errors (#390)

Code
  tidypredict_fit(model)
Condition
  Error in `parse_model()`:
  x Unable to recover the factor levels behind "a.b" and "c.d".
  i `kernlab::ksvm()` only keeps the `make.names()` form of the model matrix column names, so a level containing "." cannot be told apart from a level such as "a:b" that mangles to the same name.
  i Rename the factor levels to syntactic names before fitting.