Support xgboost on macos aarch64-apple-darwin20 (64-bit)

[aaelony-aeg]

When I try to run H2o from the Mac M1, it informs me Xgboost is not yet supported. Although I train h2o models on the cloud, it would be nice to be able to run xgboost models via H2o locally (on the Mac M1 chip) as well.

Is this on the roadmap?

Details:

library(h2o)
h2o.init()

# import the prostate dataset
df <- h2o.importFile("https://h2o-public-test-data.s3.amazonaws.com/smalldata/prostate/prostate.csv")

# convert columns to factors
df$CAPSULE <- as.factor(df$CAPSULE)
df$RACE <- as.factor(df$RACE)
df$DCAPS <- as.factor(df$DCAPS)
df$DPROS <- as.factor(df$DPROS)

# set the predictor and response columns
predictors <- c("AGE", "RACE", "VOL", "GLEASON")
response <- "CAPSULE"

# split the dataset into train and test sets
df_splits <- h2o.splitFrame(data =  df, ratios = 0.8, seed = 1234)
train <- df_splits[[1]]
test <- df_splits[[2]]

 m <- h2o.xgboost(x = predictors,
                        y = response,
                        training_frame = df)

ERROR: Unexpected HTTP Status code: 500 Server Error (url = http://localhost:54321/3/ModelBuilders/xgboost)

java.lang.IllegalStateException
 [1] "java.lang.IllegalStateException: Algorithm 'xgboost' is not registered. Available algos:
[infogram,targetencoder,deeplearning,glm,glrm,kmeans,naivebayes,pca,svd,drf,gbm,isolationforest,extendedisolationforest,aggregator,word2vec,stackedensemble,coxph,generic,gam,anovaglm,psvm,rulefit,upliftdrf,modelselection,isotonicregression,dt]"
 [2] "
hex.ModelBuilder.lambda$make$1(ModelBuilder.java:168)"


ERROR MESSAGE:

Algorithm 'xgboost' is not registered. Available algos:
[infogram,targetencoder,deeplearning,glm,glrm,kmeans,naivebayes,pca,svd,drf,gbm,isolationforest,extendedisolationforest,aggregator,word2vec,stackedensemble,coxph,generic,gam,anovaglm,psvm,rulefit,upliftdrf,modelselection,isotonicregression,dt]

> packageVersion('h2o')
[1] '3.42.0.2'

> sessionInfo()
R version 4.2.1 (2022-06-23)
Platform: aarch64-apple-darwin20 (64-bit)
Running under: macOS Ventura 13.4

Thank-you!

[wendycwong]

This one is rather involved and currently is not on the roadmap.

[aaelony-aeg]

Okay, thank-you for the quick response. (I'm probably okay running locally from a Docker container running ubuntu.)

[wendycwong]

From our XGBoost expert Adam Valenta:

Official support is from xgboost4j is not yet provided also, User decided to use docker with ubuntu. We will also need a M1 jenkyns node on our side which we did not couple of months ago