1818# mllib_regression.R: Provides methods for MLlib classification algorithms
1919# (except for tree-based algorithms) integration
2020
21+ # ' S4 class that represents an LinearSVCModel
22+ # '
23+ # ' @param jobj a Java object reference to the backing Scala LinearSVCModel
24+ # ' @export
25+ # ' @note LinearSVCModel since 2.2.0
26+ setClass ("LinearSVCModel ", representation(jobj = "jobj"))
27+
2128# ' S4 class that represents an LogisticRegressionModel
2229# '
2330# ' @param jobj a Java object reference to the backing Scala LogisticRegressionModel
@@ -39,6 +46,131 @@ setClass("MultilayerPerceptronClassificationModel", representation(jobj = "jobj"
3946# ' @note NaiveBayesModel since 2.0.0
4047setClass ("NaiveBayesModel ", representation(jobj = "jobj"))
4148
49+ # ' linear SVM Model
50+ # '
51+ # ' Fits an linear SVM model against a SparkDataFrame. It is a binary classifier, similar to svm in glmnet package
52+ # ' Users can print, make predictions on the produced model and save the model to the input path.
53+ # '
54+ # ' @param data SparkDataFrame for training.
55+ # ' @param formula A symbolic description of the model to be fitted. Currently only a few formula
56+ # ' operators are supported, including '~', '.', ':', '+', and '-'.
57+ # ' @param regParam The regularization parameter.
58+ # ' @param maxIter Maximum iteration number.
59+ # ' @param tol Convergence tolerance of iterations.
60+ # ' @param standardization Whether to standardize the training features before fitting the model. The coefficients
61+ # ' of models will be always returned on the original scale, so it will be transparent for
62+ # ' users. Note that with/without standardization, the models should be always converged
63+ # ' to the same solution when no regularization is applied.
64+ # ' @param threshold The threshold in binary classification, in range [0, 1].
65+ # ' @param weightCol The weight column name.
66+ # ' @param aggregationDepth The depth for treeAggregate (greater than or equal to 2). If the dimensions of features
67+ # ' or the number of partitions are large, this param could be adjusted to a larger size.
68+ # ' This is an expert parameter. Default value should be good for most cases.
69+ # ' @param ... additional arguments passed to the method.
70+ # ' @return \code{spark.svmLinear} returns a fitted linear SVM model.
71+ # ' @rdname spark.svmLinear
72+ # ' @aliases spark.svmLinear,SparkDataFrame,formula-method
73+ # ' @name spark.svmLinear
74+ # ' @export
75+ # ' @examples
76+ # ' \dontrun{
77+ # ' sparkR.session()
78+ # ' df <- createDataFrame(iris)
79+ # ' training <- df[df$Species %in% c("versicolor", "virginica"), ]
80+ # ' model <- spark.svmLinear(training, Species ~ ., regParam = 0.5)
81+ # ' summary <- summary(model)
82+ # '
83+ # ' # fitted values on training data
84+ # ' fitted <- predict(model, training)
85+ # '
86+ # ' # save fitted model to input path
87+ # ' path <- "path/to/model"
88+ # ' write.ml(model, path)
89+ # '
90+ # ' # can also read back the saved model and predict
91+ # ' # Note that summary deos not work on loaded model
92+ # ' savedModel <- read.ml(path)
93+ # ' summary(savedModel)
94+ # ' }
95+ # ' @note spark.svmLinear since 2.2.0
96+ setMethod ("spark.svmLinear ", signature(data = "SparkDataFrame", formula = "formula"),
97+ function (data , formula , regParam = 0.0 , maxIter = 100 , tol = 1E-6 , standardization = TRUE ,
98+ threshold = 0.0 , weightCol = NULL , aggregationDepth = 2 ) {
99+ formula <- paste(deparse(formula ), collapse = " "
100+
101+ if (! is.null(weightCol ) && weightCol == " "
102+ weightCol <- NULL
103+ } else if (! is.null(weightCol )) {
104+ weightCol <- as.character(weightCol )
105+ }
106+
107+ jobj <- callJStatic(" org.apache.spark.ml.r.LinearSVCWrapper" " fit"
108+ data @ sdf , formula , as.numeric(regParam ), as.integer(maxIter ),
109+ as.numeric(tol ), as.logical(standardization ), as.numeric(threshold ),
110+ weightCol , as.integer(aggregationDepth ))
111+ new(" LinearSVCModel" jobj = jobj )
112+ })
113+
114+ # Predicted values based on an LinearSVCModel model
115+
116+ # ' @param newData a SparkDataFrame for testing.
117+ # ' @return \code{predict} returns the predicted values based on an LinearSVCModel.
118+ # ' @rdname spark.svmLinear
119+ # ' @aliases predict,LinearSVCModel,SparkDataFrame-method
120+ # ' @export
121+ # ' @note predict(LinearSVCModel) since 2.2.0
122+ setMethod ("predict ", signature(object = "LinearSVCModel"),
123+ function (object , newData ) {
124+ predict_internal(object , newData )
125+ })
126+
127+ # Get the summary of an LinearSVCModel
128+
129+ # ' @param object an LinearSVCModel fitted by \code{spark.svmLinear}.
130+ # ' @return \code{summary} returns summary information of the fitted model, which is a list.
131+ # ' The list includes \code{coefficients} (coefficients of the fitted model),
132+ # ' \code{intercept} (intercept of the fitted model), \code{numClasses} (number of classes),
133+ # ' \code{numFeatures} (number of features).
134+ # ' @rdname spark.svmLinear
135+ # ' @aliases summary,LinearSVCModel-method
136+ # ' @export
137+ # ' @note summary(LinearSVCModel) since 2.2.0
138+ setMethod ("summary ", signature(object = "LinearSVCModel"),
139+ function (object ) {
140+ jobj <- object @ jobj
141+ features <- callJMethod(jobj , " features"
142+ labels <- callJMethod(jobj , " labels"
143+ coefficients <- callJMethod(jobj , " coefficients"
144+ nCol <- length(coefficients ) / length(features )
145+ coefficients <- matrix (unlist(coefficients ), ncol = nCol )
146+ intercept <- callJMethod(jobj , " intercept"
147+ numClasses <- callJMethod(jobj , " numClasses"
148+ numFeatures <- callJMethod(jobj , " numFeatures"
149+ if (nCol == 1 ) {
150+ colnames(coefficients ) <- c(" Estimate"
151+ } else {
152+ colnames(coefficients ) <- unlist(labels )
153+ }
154+ rownames(coefficients ) <- unlist(features )
155+ list (coefficients = coefficients , intercept = intercept ,
156+ numClasses = numClasses , numFeatures = numFeatures )
157+ })
158+
159+ # Save fitted LinearSVCModel to the input path
160+
161+ # ' @param path The directory where the model is saved.
162+ # ' @param overwrite Overwrites or not if the output path already exists. Default is FALSE
163+ # ' which means throw exception if the output path exists.
164+ # '
165+ # ' @rdname spark.svmLinear
166+ # ' @aliases write.ml,LinearSVCModel,character-method
167+ # ' @export
168+ # ' @note write.ml(LogisticRegression, character) since 2.2.0
169+ setMethod ("write.ml ", signature(object = "LinearSVCModel", path = "character"),
170+ function (object , path , overwrite = FALSE ) {
171+ write_internal(object , path , overwrite )
172+ })
173+
42174# ' Logistic Regression Model
43175# '
44176# ' Fits an logistic regression model against a SparkDataFrame. It supports "binomial": Binary logistic regression
0 commit comments