Skip to content

Stratified sampling #1336

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 15 commits into from
Feb 29, 2024
Merged

Stratified sampling #1336

Changes from 2 commits
Commits
Show all changes
15 commits
Select commit Hold shift + click to select a range
0574574
handle model deploy when no metrics to compare
ChuckHend Jan 17, 2024
667f68e
better warn msg
ChuckHend Jan 17, 2024
d0ff725
fix first run case
ChuckHend Jan 17, 2024
accaab0
impl stratified
ChuckHend Jan 17, 2024
5831025
handle case where exists has no metrics
ChuckHend Jan 17, 2024
647a75b
Merge branch 'master' into chuck/stratifiedSampling
ChuckHend Jan 17, 2024
7710b13
change default samping to stratified
ChuckHend Jan 17, 2024
3083efa
Merge branch 'master' of https://github.com/ChuckHend/postgresml
ChuckHend Jan 18, 2024
8d089f2
Merge branch 'master' into chuck/stratifiedSampling
ChuckHend Jan 18, 2024
b35d695
Merge pull request #1 from ChuckHend/chuck/stratifiedSampling
ChuckHend Jan 18, 2024
9b2c44a
no rando when already materialized
ChuckHend Jan 19, 2024
ec699b9
Merge branch 'master' into master
ChuckHend Jan 19, 2024
489bce1
update enum and function signatures
ChuckHend Jan 19, 2024
7a83a13
merge conflict
ChuckHend Feb 29, 2024
0af75a8
add upgrade test
ChuckHend Feb 29, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 21 additions & 15 deletions pgml-extension/src/api.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -278,23 +278,29 @@ fn train_joint(
"Comparing to deployed model {}: {:?}",
default_target_metric, deployed_metric
);
if let (Some(deployed_metric_value), Some(new_metric_value)) = (
deployed_metric,
new_metrics.get(&default_target_metric).and_then(|v| v.as_f64()),
) {
if project.task.value_is_better(deployed_metric_value, new_metric_value) {
warning!(
"New model's {} is not better than current model. New: {}, Current {}",
&project.task.default_target_metric(),
new_metric_value,
deployed_metric_value
);
let new_metric = new_metrics.get(&default_target_metric).and_then(|v| v.as_f64());

match (deployed_metric, new_metric) {
(Some(deployed), Some(new)) => {
// only compare metrics when both new and old model have metrics to compare
if project.task.value_is_better(deployed, new) {
warning!(
"New model's {} is not better than current model. New: {}, Current {}",
&default_target_metric,
new,
deployed
);
deploy = false;
}
}
(None, None) => {
warning!("No metrics available for both deployed and new model. Deploying new model.")
}
(Some(_deployed), None) => {
warning!("No metrics for new model. Retaining old model.");
deploy = false;
}
} else {
warning!("Failed to retrieve or parse deployed/new metrics for {}. Ensure train/test split results in both positive and negative label records.",
&project.task.default_target_metric());
deploy = false;
(None, Some(_new)) => warning!("No metrics for deployed model. Deploying new model."),
}
} else {
warning!("Failed to parse deployed model metrics. Check data types of model metadata on pgml.models.metrics");
Expand Down