Question

我已经搜索了不同的门户网站，甚至在h2o集合文档中，所有我都有集合示例，仅用于分类问题二进制本质，但没有一个示例显示如何实现一般堆叠或h2o集合以解决r中的简单回归问题

我请求任何人请分享有关如何实施h2o整体或仅为R中的回归问题堆叠的工作代码

OR

简单的集合仅适用于R中的回归。

只想知道如何为不同权重的回归实施集合/堆叠。

Answer 1

以下是为R中的回归问题（预测年龄）构建堆叠集合的示例：

library('h2o')
h2o.init()

files3 = "http://h2o-public-test-data.s3.amazonaws.com/smalldata/prostate/prostate.csv"
col_types <- c("Numeric","Numeric","Numeric","Enum","Enum","Numeric","Numeric","Numeric","Numeric")
dat <- h2o.importFile(files3,destination_frame = "prostate.hex",col.types = col_types)
ss <- h2o.splitFrame(dat, ratios = 0.8, seed = 1)
train <- ss[[1]]
test <- ss[[2]]

x <- c("CAPSULE","GLEASON","RACE","DPROS","DCAPS","PSA","VOL")
y <- "AGE"
nfolds <- 5


# Train & Cross-validate a GBM
my_gbm <- h2o.gbm(x = x, 
                  y = y, 
                  training_frame = train, 
                  distribution = "gaussian",
                  max_depth = 3,
                  learn_rate = 0.2,
                  nfolds = nfolds, 
                  fold_assignment = "Modulo",
                  keep_cross_validation_predictions = TRUE,
                  seed = 1)

# Train & Cross-validate a RF
my_rf <- h2o.randomForest(x = x,
                          y = y, 
                          training_frame = train, 
                          ntrees = 30, 
                          nfolds = nfolds, 
                          fold_assignment = "Modulo",
                          keep_cross_validation_predictions = TRUE,
                          seed = 1)


# Train & Cross-validate a extremely-randomized RF
my_xrf <- h2o.randomForest(x = x,
                           y = y, 
                           training_frame = train, 
                           ntrees = 50,
                           histogram_type = "Random",
                           nfolds = nfolds, 
                           fold_assignment = "Modulo",
                           keep_cross_validation_predictions = TRUE,
                           seed = 1)

# Train a stacked ensemble using the models above
stack <- h2o.stackedEnsemble(x = x, 
                             y = y, 
                             training_frame = train,
                             validation_frame = test,  #also test that validation_frame is working
                             model_id = "my_ensemble_gaussian", 
                             base_models = list(my_gbm@model_id, my_rf@model_id, my_xrf@model_id))

# predict
pred <- h2o.predict(stack, newdata = test)

Answer 2

我的书（使用H2O的实用机器学习）中的堆叠整体示例是回归（在建筑能量数据集上）。： - ）

但是，如果你认为你已经用H2O耗尽了所有文档，请尝试在github上搜索源代码。这是他们对堆叠整体回归的单元测试：

https://github.com/h2oai/h2o-3/blob/master/h2o-r/tests/testdir_algos/stackedensemble/runit_stackedensemble_gaussian.R

r中的h2o集合实现的代码r

2 个答案: