Repository: spark Updated Branches: refs/heads/master 74b47845e -> 55aa4da28
[SPARK-21622][ML][SPARKR] Support offset in SparkR GLM ## What changes were proposed in this pull request? Support offset in SparkR GLM #16699 Author: actuaryzhang <actuaryzhan...@gmail.com> Closes #18831 from actuaryzhang/sparkROffset. Project: http://git-wip-us.apache.org/repos/asf/spark/repo Commit: http://git-wip-us.apache.org/repos/asf/spark/commit/55aa4da2 Tree: http://git-wip-us.apache.org/repos/asf/spark/tree/55aa4da2 Diff: http://git-wip-us.apache.org/repos/asf/spark/diff/55aa4da2 Branch: refs/heads/master Commit: 55aa4da285ef0f2fa43852f2bbcffacc7ddefdbf Parents: 74b4784 Author: actuaryzhang <actuaryzhan...@gmail.com> Authored: Sun Aug 6 15:14:12 2017 -0700 Committer: Felix Cheung <felixche...@apache.org> Committed: Sun Aug 6 15:14:12 2017 -0700 ---------------------------------------------------------------------- R/pkg/R/mllib_regression.R | 22 ++++++++++++++++---- R/pkg/tests/fulltests/test_mllib_regression.R | 8 +++++++ .../r/GeneralizedLinearRegressionWrapper.scala | 4 +++- 3 files changed, 29 insertions(+), 5 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/spark/blob/55aa4da2/R/pkg/R/mllib_regression.R ---------------------------------------------------------------------- diff --git a/R/pkg/R/mllib_regression.R b/R/pkg/R/mllib_regression.R index 9ecd887..ebaeae9 100644 --- a/R/pkg/R/mllib_regression.R +++ b/R/pkg/R/mllib_regression.R @@ -76,6 +76,8 @@ setClass("IsotonicRegressionModel", representation(jobj = "jobj")) #' "frequencyDesc", "frequencyAsc", "alphabetDesc", and "alphabetAsc". #' The default value is "frequencyDesc". When the ordering is set to #' "alphabetDesc", this drops the same category as R when encoding strings. +#' @param offsetCol the offset column name. If this is not set or empty, we treat all instance offsets +#' as 0.0. The feature specified as offset has a constant coefficient of 1.0. #' @param ... additional arguments passed to the method. #' @aliases spark.glm,SparkDataFrame,formula-method #' @return \code{spark.glm} returns a fitted generalized linear model. @@ -127,7 +129,8 @@ setMethod("spark.glm", signature(data = "SparkDataFrame", formula = "formula"), function(data, formula, family = gaussian, tol = 1e-6, maxIter = 25, weightCol = NULL, regParam = 0.0, var.power = 0.0, link.power = 1.0 - var.power, stringIndexerOrderType = c("frequencyDesc", "frequencyAsc", - "alphabetDesc", "alphabetAsc")) { + "alphabetDesc", "alphabetAsc"), + offsetCol = NULL) { stringIndexerOrderType <- match.arg(stringIndexerOrderType) if (is.character(family)) { @@ -159,12 +162,19 @@ setMethod("spark.glm", signature(data = "SparkDataFrame", formula = "formula"), weightCol <- as.character(weightCol) } + if (!is.null(offsetCol)) { + offsetCol <- as.character(offsetCol) + if (nchar(offsetCol) == 0) { + offsetCol <- NULL + } + } + # For known families, Gamma is upper-cased jobj <- callJStatic("org.apache.spark.ml.r.GeneralizedLinearRegressionWrapper", "fit", formula, data@sdf, tolower(family$family), family$link, tol, as.integer(maxIter), weightCol, regParam, as.double(var.power), as.double(link.power), - stringIndexerOrderType) + stringIndexerOrderType, offsetCol) new("GeneralizedLinearRegressionModel", jobj = jobj) }) @@ -192,6 +202,8 @@ setMethod("spark.glm", signature(data = "SparkDataFrame", formula = "formula"), #' "frequencyDesc", "frequencyAsc", "alphabetDesc", and "alphabetAsc". #' The default value is "frequencyDesc". When the ordering is set to #' "alphabetDesc", this drops the same category as R when encoding strings. +#' @param offsetCol the offset column name. If this is not set or empty, we treat all instance offsets +#' as 0.0. The feature specified as offset has a constant coefficient of 1.0. #' @return \code{glm} returns a fitted generalized linear model. #' @rdname glm #' @export @@ -209,10 +221,12 @@ setMethod("glm", signature(formula = "formula", family = "ANY", data = "SparkDat function(formula, family = gaussian, data, epsilon = 1e-6, maxit = 25, weightCol = NULL, var.power = 0.0, link.power = 1.0 - var.power, stringIndexerOrderType = c("frequencyDesc", "frequencyAsc", - "alphabetDesc", "alphabetAsc")) { + "alphabetDesc", "alphabetAsc"), + offsetCol = NULL) { spark.glm(data, formula, family, tol = epsilon, maxIter = maxit, weightCol = weightCol, var.power = var.power, link.power = link.power, - stringIndexerOrderType = stringIndexerOrderType) + stringIndexerOrderType = stringIndexerOrderType, + offsetCol = offsetCol) }) # Returns the summary of a model produced by glm() or spark.glm(), similarly to R's summary(). http://git-wip-us.apache.org/repos/asf/spark/blob/55aa4da2/R/pkg/tests/fulltests/test_mllib_regression.R ---------------------------------------------------------------------- diff --git a/R/pkg/tests/fulltests/test_mllib_regression.R b/R/pkg/tests/fulltests/test_mllib_regression.R index 6b72a09..23daca7 100644 --- a/R/pkg/tests/fulltests/test_mllib_regression.R +++ b/R/pkg/tests/fulltests/test_mllib_regression.R @@ -173,6 +173,14 @@ test_that("spark.glm summary", { expect_equal(stats$df.residual, rStats$df.residual) expect_equal(stats$aic, rStats$aic) + # Test spark.glm works with offset + training <- suppressWarnings(createDataFrame(iris)) + stats <- summary(spark.glm(training, Sepal_Width ~ Sepal_Length + Species, + family = poisson(), offsetCol = "Petal_Length")) + rStats <- suppressWarnings(summary(glm(Sepal.Width ~ Sepal.Length + Species, + data = iris, family = poisson(), offset = iris$Petal.Length))) + expect_true(all(abs(rStats$coefficients - stats$coefficients) < 1e-3)) + # Test summary works on base GLM models baseModel <- stats::glm(Sepal.Width ~ Sepal.Length + Species, data = iris) baseSummary <- summary(baseModel) http://git-wip-us.apache.org/repos/asf/spark/blob/55aa4da2/mllib/src/main/scala/org/apache/spark/ml/r/GeneralizedLinearRegressionWrapper.scala ---------------------------------------------------------------------- diff --git a/mllib/src/main/scala/org/apache/spark/ml/r/GeneralizedLinearRegressionWrapper.scala b/mllib/src/main/scala/org/apache/spark/ml/r/GeneralizedLinearRegressionWrapper.scala index 176a6cf..64575b0 100644 --- a/mllib/src/main/scala/org/apache/spark/ml/r/GeneralizedLinearRegressionWrapper.scala +++ b/mllib/src/main/scala/org/apache/spark/ml/r/GeneralizedLinearRegressionWrapper.scala @@ -77,7 +77,8 @@ private[r] object GeneralizedLinearRegressionWrapper regParam: Double, variancePower: Double, linkPower: Double, - stringIndexerOrderType: String): GeneralizedLinearRegressionWrapper = { + stringIndexerOrderType: String, + offsetCol: String): GeneralizedLinearRegressionWrapper = { // scalastyle:on val rFormula = new RFormula().setFormula(formula) .setStringIndexerOrderType(stringIndexerOrderType) @@ -99,6 +100,7 @@ private[r] object GeneralizedLinearRegressionWrapper glr.setLink(link) } if (weightCol != null) glr.setWeightCol(weightCol) + if (offsetCol != null) glr.setOffsetCol(offsetCol) val pipeline = new Pipeline() .setStages(Array(rFormulaModel, glr)) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@spark.apache.org For additional commands, e-mail: commits-h...@spark.apache.org
