Skip to content

Handling vetiver objects

A vetiver_model() collects the information needed to store, version, and deploy a trained model.

vetiver_model() new_vetiver_model()
Create a vetiver object for deployment of a trained model
vetiver_pin_write() vetiver_pin_read()
Read and write a trained model to a board of models
vetiver_api() vetiver_pr_post() vetiver_pr_docs()
Create a Plumber API to predict with a deployable vetiver_model() object
Write a deployable Plumber file for a vetiver model
Write a Dockerfile for a vetiver model
Deploy a vetiver model API to RStudio Connect
Create an RStudio Connect bundle for a vetiver model API
Create a model API endpoint object for prediction
Post new data to a deployed model API endpoint and return predictions
Post new data to a deployed model API endpoint and augment with predictions

Monitoring deployed models

Monitor a deployed vetiver_model() with a dashboard.

Aggregate model metrics over time for monitoring
Update model metrics over time for monitoring
Plot model metrics over time for monitoring
vetiver_dashboard() get_vetiver_dashboard_pins() pin_example_kc_housing_model()
R Markdown format for model monitoring dashboards

Developer functions

These functions are helpful for developers extending vetiver for other types of models.

api_spec() glue_spec_summary()
Update the OpenAPI specification using model metadata
attach_pkgs() load_pkgs()
Fully attach or load packages for making model predictions
handler_startup() handler_predict()
Model handler functions for API endpoint
Identify data types for each column in an input data prototype
vetiver_create_description() vetiver_prepare_model()
Model constructor methods
vetiver_meta() vetiver_create_meta()
Metadata constructors for vetiver_model() object
vetiver_ptype() vetiver_create_ptype()
Create a vetiver input data prototype
Convert new data at prediction time using input data prototype