This is an automated email from the ASF dual-hosted git repository.
ssiddiqi pushed a change to branch main
in repository https://gitbox.apache.org/repos/asf/systemds.git.
from 62a66b0 [MINOR] Performance and cleanup winsorizeApply built-in
function
add bfab2a7 [MINOR] Performance improvements in cleaning pipelines -
This commit changes the outer for loop of bandit::run_with_hyperparam to parfor
alongside with previous improvements i.e. winsorizeApply, this commit
brings down the execution time on EEG dataset from 14000 sec to 2200 sec.
- This commit also simplifies the sampling function inside utils.dml
No new revisions were added by this update.
Summary of changes:
scripts/builtin/applyAndEvaluate.dml | 7 +-
scripts/builtin/bandit.dml | 33 +--
scripts/builtin/executePipeline.dml | 2 +-
scripts/builtin/mice.dml | 2 +-
scripts/builtin/topk_cleaning.dml | 62 +++--
scripts/pipelines/scripts/enumerateLogical.dml | 25 +-
scripts/pipelines/scripts/utils.dml | 274 ++++++---------------
.../BuiltinTopkCleaningClassificationTest.java | 2 +-
.../pipelines/BuiltinTopkEvaluateTest.java | 3 +-
.../functions/pipelines/applyEvaluateTest.dml | 1 -
.../intermediates/classification/applyFunc.csv | 4 +-
.../intermediates/classification/bestAcc.csv | 4 +-
.../pipelines/intermediates/classification/hp.csv | 6 +-
.../pipelines/intermediates/classification/pip.csv | 4 +-
14 files changed, 150 insertions(+), 279 deletions(-)
