how to check the value of anomoly detction i mean the standa Elementary Community #ask-elementary-ai

Join Slack

how to check the value of anomoly detction i mean ...

# ask-elementary-ai

shobia

11/24/2025, 7:11 AM

how to check the value of anomoly detction i mean the standard deviation in the schema tables

kapa.ai

11/24/2025, 7:11 AM

A message from kapa.ai

kapa.ai

11/24/2025, 7:11 AM

You can inspect the underlying anomaly calculations directly in Elementary’s warehouse tables/views. ### 1. Where standard deviation is stored For anomaly tests, Elementary first aggregates metrics into: -

data_monitoring_metrics

– stores the raw metric values per bucket (e.g. per day/hour) for each anomaly test. [[DBT artifacts](https://slack.com/archives/C02CTC89LAX/p1663047264555179); [Tables overview](https://slack.com/archives/C02CTC89LAX/p1666376716952649)] Standard deviation and other training stats (avg, training set size, etc.) are then calculated in: -

metrics_anomaly_score

– a model/view that reads from

data_monitoring_metrics

and computes: -

training_avg

training_stddev

training_set_size

anomaly_score

[[Anomaly scores SQL](https://slack.com/archives/C02CTC89LAX/p1687939167177509); [Tables overview](https://slack.com/archives/C02CTC89LAX/p1666376716952649)] Per‑test temporary tables ending with

__anomaly_scores

in the

elementary__tests

schema also contain

training_stddev

training_avg

min_metric_value

max_metric_value

, etc. [[Per‑test scores](https://slack.com/archives/C02CTC89LAX/p1687939167177509)] ### 2. Example queries a. See all training stddev values from the central model:

Copy code

sql SELECT   full_table_name,   column_name,   metric_name,   bucket_start,   bucket_end,   metric_value,   training_avg,   training_stddev,   training_set_size,   anomaly_score FROM <db>.<elementary_schema>.metrics_anomaly_score ORDER BY bucket_end DESC;

b. Check the raw metrics going into the calculation:

Copy code

sql SELECT   full_table_name,   column_name,   metric_name,   bucket_start,   bucket_end,   metric_value,   updated_at FROM <db>.<elementary_schema>.data_monitoring_metrics ORDER BY bucket_end DESC;

c. Look at a specific anomaly test’s own
__anomaly_scores
table (if you know its name):

Copy code

sql SELECT   anomaly_score,   anomaly_score_threshold,   training_avg,   training_stddev,   min_metric_value,   max_metric_value,   bucket_start,   metric_value FROM <db>.<elementary_tests_schema>.<some_test_name__anomaly_scores> ORDER BY bucket_start DESC;

In all cases,

training_stddev

is the standard deviation Elementary uses to compute the z‑score–based

anomaly_score

. [[Anomaly method](https://docs.elementary-data.com/data-tests/data-anomaly-detection); [Anomaly scores SQL](https://slack.com/archives/C02CTC89LAX/p1687939167177509)]

27 Views

Open in Slack

Previous Next