Celes Renata
34 lines
1.4 KiB
SQL
34 lines
1.4 KiB
SQL
-- Analytical fact table: model_performance
|
|
-- Tracks extraction model performance for Trino/Superset dashboards.
|
|
-- Partitioned by dt and model_name on MinIO.
|
|
-- Path: s3://stonks-lakehouse/warehouse/model_performance/dt={yyyy-mm-dd}/model_name={name}/part-*.parquet
|
|
-- Requirements: 12.1, 12.2
|
|
|
|
CREATE TABLE IF NOT EXISTS lakehouse.stonks.model_performance (
|
|
document_id VARCHAR,
|
|
ticker VARCHAR,
|
|
model_name VARCHAR,
|
|
prompt_version VARCHAR,
|
|
schema_version VARCHAR,
|
|
success BOOLEAN,
|
|
attempt_count INTEGER,
|
|
total_duration_ms INTEGER,
|
|
first_attempt_duration_ms INTEGER,
|
|
final_attempt_duration_ms INTEGER,
|
|
confidence DOUBLE,
|
|
validation_status VARCHAR,
|
|
validation_error_count INTEGER,
|
|
validation_warning_count INTEGER,
|
|
retry_count INTEGER,
|
|
input_token_estimate INTEGER,
|
|
output_token_estimate INTEGER,
|
|
company_count INTEGER,
|
|
recorded_at TIMESTAMP(6) WITH TIME ZONE,
|
|
dt DATE,
|
|
model_version VARCHAR
|
|
) WITH (
|
|
format = 'PARQUET',
|
|
partitioned_by = ARRAY['dt', 'model_version'],
|
|
external_location = 's3a://stonks-lakehouse/warehouse/model_performance/'
|
|
);
|