Export a Spark pipeline for serving

ml_write_to_bundle_transformed

Description

This functions serializes a Spark pipeline model into an MLeap bundle.

Usage

ml_write_to_bundle_transformed(x, transformed_dataset, path, overwrite = FALSE)

Arguments

Arguments	Description
x	A Spark ML Pipeline Model object.
transformed_dataset	A Spark data frame created by the ML Pipeline Model (x)
path	Where to save the bundle.
overwrite	Whether to overwrite an existing file, defaults to FALSE.

Examples

library(sparklyr)

sc <- spark_connect(master = "local")

mtcars_tbl <- copy_to(sc, mtcars, overwrite = TRUE)

pipeline <- ml_pipeline(sc) %>%
  ft_binarizer("hp", "big_hp", threshold = 100) %>%
  ft_vector_assembler(c("big_hp", "wt", "qsec"), "features") %>%
  ml_gbt_regressor(label_col = "mpg")

pipeline_model <- ml_fit(pipeline, mtcars_tbl)

preds <- ml_transform(pipeline_model, mtcars_tbl)

model_path <- file.path(tempdir(), "mtcars_model.zip")

ml_write_bundle(
  x = pipeline_model,
  transformed_dataset = preds,
  path = model_path,
  overwrite = TRUE
)