Variable importance heatmap for AutoML inside h2o.explain shows too many features

[exalate-issue-sync]

I think we forgot to limit the number of features shown in the heatmap to the top 20? Here's an image where it's showing way more (this dataset has 4k features – [guillermo|https://www.openml.org/d/41159]).

The {{h2o.varimp_heatmap()}} has default {{top_n = 20}} but it doesn’t look like it’s working.

!Screen Shot 2021-08-20 at 5.10.52 PM.png|width=200,height=183!

{noformat}library(OpenML)
library(h2o)

data_id <- 41159
#guillermo: https://www.openml.org/d/41159 target: "class"

h2o.init()

omld <- getOMLDataSet(data.id = data_id)
y <- omld$target.features
data <- as.h2o(omld$data)
data_name <- omld$desc$name

ss <- h2o.splitFrame(data, ratios = 0.8, seed = 1)
train <- ss[[1]]
test <- ss[[2]]
dim(train)

aml_fullt <- h2o.automl(y = y,
training_frame = train,
leaderboard_frame = test,
max_runtime_secs = 3600,
project_name = paste0(data_name, "_fullt"),
seed = 1)

lb_fullt <- aml_fullt@leaderboard
print(lb_fullt, n = nrow(lb_fullt))
aml_fullt@training_info$duration_secs

h2o.explain(aml_fullt, test) #plot is from here{noformat}

[exalate-issue-sync]

Tomas Fryda commented: {{top_n}} refers to number of models not features.

[exalate-issue-sync]

Erin LeDell commented: Duplicate of https://h2oai.atlassian.net/browse/PUBDEV-8014

[h2o-ops-ro]

JIRA Issue Details

Jira Issue: PUBDEV-8289
Assignee: Tomas Fryda
Reporter: Erin LeDell
State: Closed
Fix Version: 3.34.0.1
Attachments: Available (Count: 1)
Development PRs: N/A

[h2o-ops-ro]

Attachments From Jira

Attachment Name: Screen Shot 2021-08-20 at 5.10.52 PM.png
Attached By: Erin LeDell
File Link:https://h2o-3-jira-github-migration.s3.amazonaws.com/PUBDEV-8289/Screen Shot 2021-08-20 at 5.10.52 PM.png