Error: object 'h2o.targetencoder' not found

tlg265 · September 25, 2019, 6:38am

I want to use: h2o.targetencoder function, then I do:

> install.packages("h2o")
> library("h2o")
> h2o.targetencoder

but I get the following error:

Error: object 'h2o.targetencoder' not found

At the same time, I see that function is defined here:

github.com

h2oai/h2o-3/blob/jenkins-rel-yau-5/h2o-r/h2o-package/R/targetencoder.R

# This file is auto-generated by h2o-3/h2o-bindings/bin/gen_R.py
# Copyright 2016 H2O.ai;  Apache License Version 2.0 (see LICENSE for details) 
#'
# -------------------------- Target Encoder -------------------------- #
#'
#' Transformation of a categorical variable with a mean value of the target variable
#'
#' @param x (Optional) A vector containing the names or indices of the predictor variables to use in building the model.
#'        If x is missing, then all columns except y are used.
#' @param y The name or column index of the response variable in the data. 
#'        The response must be either a numeric or a categorical/factor variable. 
#'        If the response is numeric, then a regression model will be trained, otherwise it will train a classification model.
#' @param training_frame Id of the training data frame.
#' @param blending \code{Logical}. Blending enabled/disabled Defaults to FALSE.
#' @param k Inflection point. Used for blending (if enabled). Blending is to be enabled separately using the 'blending'
#'        parameter. Defaults to 20.
#' @param f Smooothing. Used for blending (if enabled). Blending is to be enabled separately using the 'blending'
#'        parameter. Defaults to 10.
#' @param data_leakage_handling Data leakage handling strategy. Default to None. Must be one of: "None", "KFold", "LeaveOneOut". Defaults to
#'        None.

This file has been truncated. show original

Any idea on how can I use function: h2o.targetencoder?

Thanks!

andresrcs · September 25, 2019, 3:25pm

That is a new function only included in the development version of h2o you can install it from this repo.

install.packages("h2o", type="source", repos="http://h2o-release.s3.amazonaws.com/h2o/master/4799/R")

tlg265 · September 25, 2019, 3:33pm

thank you @andresrcs!

tlg265 · September 27, 2019, 2:55pm

Just taking a note here...

My goal was to do Target Encoding when the target variable is continuous.

I installed the latest package (as of now) from GitHub repository, but got the following error:

java.lang.IllegalStateException: `target` must be a binary categorical vector. We do not support multi-class and continuos target case for now

java.lang.IllegalStateException: `target` must be a binary categorical vector. We do not support multi-class and continuos target case for now
	at ai.h2o.targetencoding.TargetEncoder.ensureTargetColumnIsBinaryCategorical(TargetEncoder.java:188)
	at ai.h2o.targetencoding.TargetEncoder.prepareEncodingMap(TargetEncoder.java:137)
	at ai.h2o.targetencoding.TargetEncoder.prepareEncodingMap(TargetEncoder.java:198)
	at ai.h2o.targetencoding.TargetEncoderBuilder$TargetEncoderDriver.computeImpl(TargetEncoderBuilder.java:45)
	at hex.ModelBuilder$Driver.compute2(ModelBuilder.java:222)
	at water.H2O$H2OCountedCompleter.compute(H2O.java:1417)
	at jsr166y.CountedCompleter.exec(CountedCompleter.java:468)
	at jsr166y.ForkJoinTask.doExec(ForkJoinTask.java:263)
	at jsr166y.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:974)
	at jsr166y.ForkJoinPool.runWorker(ForkJoinPool.java:1477)
	at jsr166y.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:104)

Error: java.lang.IllegalStateException: `target` must be a binary categorical vector. We do not support multi-class and continuos target case for now

It seems that the this is not an option for me right now for Target Encoding when target continuous variable.

Thanks!

valeri · September 27, 2019, 7:34pm

Hi @tlg265,

I am wondering what you are trying to achieve by trying to encode a continuous variable. For a categorical variable like c('cat', 'dog') one can encode this as c(1,2) with the category cat encoded as 1 and dog - as 2. But for a continuous numeric variable I am not sure if/what encoding can be used (for).

tlg265 · October 2, 2019, 5:12am

Hi @valeri, I was not trying to target encode a predictor continuous variable. What I said above is:
"this is not an option for me right now for Target Encoding when target continuous variable"

what it is continue is the target variable, but variable: animal (which will be encoded) is categorical.

Thanks!

valeri · October 2, 2019, 6:37am

Thanks @tlg265, I think I understand but I am not sure - if the operation is called targetencoding and the target is continuous, then isn't that what it is trying to encode? In my comment, I actually meant it in general, regardless if the continuous variable is a target or predictor - I still don't see sense in encoding it, unless by encoding we mean binning it somehow.

jcblum · October 2, 2019, 10:07am

Sometimes this procedure is called "target-based encoding", which is perhaps a clearer name. What's being encoded is not the continuous "target" — instead, a continuous (or categorical) "target" variable is being used to encode a categorical variable. See:

Encoding (simple overview)
Why you should try Mean Encoding. One very common step in any feature… | by Miguel José Monteiro | Towards Data Science
A preprocessing scheme for high-cardinality categorical attributes in classification and prediction problems | ACM SIGKDD Explorations Newsletter (Micci-Barreca 2001, ACM SIGKDD Explorations Newsletter)

valeri · October 2, 2019, 10:11am

Oh ... thanks @jcblum - now I get it

system · October 9, 2019, 10:11am

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.