ml_prepare_dataframe
Prepare a Spark DataFrame for Spark ML Routines
This routine prepares a Spark DataFrame for use by Spark ML routines.
Usage
ml_prepare_dataframe(x, features, response = NULL, ...,
ml.options = ml_options(), envir = new.env(parent = emptyenv()))
Arguments
- x
An object coercable to a Spark DataFrame (typically, a
tbl_spark
).- features
The name of features (terms) to use for the model fit.
- response
The name of the response vector (as a length-one character vector), or a formula, giving a symbolic description of the model to be fitted. When
response
is a formula, it is used in preference to other parameters to set theresponse
,features
, andintercept
parameters (if available). Currently, only simple linear combinations of existing parameters is supposed; e.g.response ~ feature1 + feature2 + ...
. The intercept term can be omitted by using- 1
in the model fit.- ...
Optional arguments. The
data
argument can be used to specify the data to be used whenx
is a formula; this allows calls of the formml_linear_regression(y ~ x, data = tbl)
, and is especially useful in conjunction withdo
.- ml.options
Optional arguments, used to affect the model generated. See
ml_options
for more details.- envir
An R environment -- when supplied, it will be filled with metadata describing the transformations that have taken place.
Details
Spark DataFrames are prepared through the following transformations:
All specified columns are transformed into a numeric data type (using a simple cast for integer / logical columns, and
ft_string_indexer
for strings),The
ft_vector_assembler
is used to combine the specified features into a single 'feature' vector, suitable for use with Spark ML routines.
After calling this function, the envir
environment (when supplied)
will be populated with a set of variables:
features : |
The name of the generated features vector. |
response : |
The name of the generated response vector. |
Examples
# NOT RUN {
# example of how 'ml_prepare_dataframe' might be used to invoke
# Spark's LinearRegression routine from the 'ml' package
envir <- new.env(parent = emptyenv())
tdf <- ml_prepare_dataframe(df, features, response, envir = envir)
lr <- invoke_new(
sc,
"org.apache.spark.ml.regression.LinearRegression"
)
# use generated 'features', 'response' vector names in model fit
model <- lr %>%
invoke("setFeaturesCol", envir$features) %>%
invoke("setLabelCol", envir$response)
# }