Skip to contents

spark.fmClassifier fits a factorization classification model against a SparkDataFrame. Users can call summary to print a summary of the fitted model, predict to make predictions on new data, and write.ml/read.ml to save/load fitted models. Only categorical data is supported.

Usage

spark.fmClassifier(data, formula, ...)

# S4 method for SparkDataFrame,formula
spark.fmClassifier(
  data,
  formula,
  factorSize = 8,
  fitLinear = TRUE,
  regParam = 0,
  miniBatchFraction = 1,
  initStd = 0.01,
  maxIter = 100,
  stepSize = 1,
  tol = 1e-06,
  solver = c("adamW", "gd"),
  thresholds = NULL,
  seed = NULL,
  handleInvalid = c("error", "keep", "skip")
)

# S4 method for FMClassificationModel
summary(object)

# S4 method for FMClassificationModel
predict(object, newData)

# S4 method for FMClassificationModel,character
write.ml(object, path, overwrite = FALSE)

Arguments

data

a SparkDataFrame of observations and labels for model fitting.

formula

a symbolic description of the model to be fitted. Currently only a few formula operators are supported, including '~', '.', ':', '+', and '-'.

...

additional arguments passed to the method.

factorSize

dimensionality of the factors.

fitLinear

whether to fit linear term. # TODO Can we express this with formula?

regParam

the regularization parameter.

miniBatchFraction

the mini-batch fraction parameter.

initStd

the standard deviation of initial coefficients.

maxIter

maximum iteration number.

stepSize

stepSize parameter.

tol

convergence tolerance of iterations.

solver

solver parameter, supported options: "gd" (minibatch gradient descent) or "adamW".

thresholds

in binary classification, in range [0, 1]. If the estimated probability of class label 1 is > threshold, then predict 1, else 0. A high threshold encourages the model to predict 0 more often; a low threshold encourages the model to predict 1 more often. Note: Setting this with threshold p is equivalent to setting thresholds c(1-p, p).

seed

seed parameter for weights initialization.

handleInvalid

How to handle invalid data (unseen labels or NULL values) in features and label column of string type. Supported options: "skip" (filter out rows with invalid data), "error" (throw an error), "keep" (put invalid data in a special additional bucket, at index numLabels). Default is "error".

object

a FM Classification model fitted by spark.fmClassifier.

newData

a SparkDataFrame for testing.

path

The directory where the model is saved.

overwrite

Overwrites or not if the output path already exists. Default is FALSE which means throw exception if the output path exists.

Value

spark.fmClassifier returns a fitted Factorization Machines Classification Model. summary returns summary information of the fitted model, which is a list. predict returns the predicted values based on a FM Classification model.

Note

spark.fmClassifier since 3.1.0

summary(FMClassificationModel) since 3.1.0

predict(FMClassificationModel) since 3.1.0

write.ml(FMClassificationModel, character) since 3.1.0

See also

Examples

if (FALSE) {
df <- read.df("data/mllib/sample_binary_classification_data.txt", source = "libsvm")

# fit Factorization Machines Classification Model
model <- spark.fmClassifier(
           df, label ~ features,
           regParam = 0.01, maxIter = 10, fitLinear = TRUE
         )

# get the summary of the model
summary(model)

# make predictions
predictions <- predict(model, df)

# save and load the model
path <- "path/to/model"
write.ml(model, path)
savedModel <- read.ml(path)
summary(savedModel)
}