library(mleap)
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
)
Export a Spark pipeline for serving
R/write-bundle.R
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. |