Feature Request: Expand supported model types and engines

[EmilHvitfeldt]

Based on analysis of the parsnip ecosystem, there are several model types and engines that could be supported by tidypredict for in-database prediction. All of these use SQL-compatible operations already employed by the package.

If there's a specific model or engine you'd like to see supported, please comment below!

Already Supported Models (new engines)

• linear_reg()
  • lm
  • glmnet
  • glm
  • quantreg
• logistic_reg()
  • glm
  • glmnet
  • LiblineaR
• rand_forest()
  • randomForest
  • ranger
  • partykit
  • grf
  • aorsf
• decision_tree()
  • partykit
  • rpart
  • C5.0
• boost_tree()
  • xgboost
  • lightgbm
  • catboost
  • C5.0
  • h2o
  • mboost
• mars()
  • earth
• cubist_rules()
  • Cubist

Proposed Models

• C5_rules()
  • C5.0
• svm_linear()
  • LiblineaR
  • kernlab
• multinom_reg()
  • nnet
  • glmnet
• rule_fit()
  • xrf
  • h2o
• discrim_linear()
  • MASS
  • mda
  • sda
  • sparsediscrim
• pls()
  • mixOmics
• null_model()
  • parsnip
• naive_Bayes()
  • klaR
  • naivebayes
• mlp() — small hidden layers only
  • nnet
• bart()
  • dbarts
• bag_tree()
  • rpart
  • C5.0
• discrim_quad()
  • MASS
• proportional_hazards() — linear predictor/hazard ratio only
  • survival
  • glmnet