2024-02-27 14:29:22 -06:00
|
|
|
-- Create staging table for results
|
|
|
|
|
CREATE TABLE IF NOT EXISTS STG.TRAINING_RESULTS
|
|
|
|
|
(
|
|
|
|
|
CONTRACT_NAME VARCHAR,
|
|
|
|
|
CONTRACT_ID NUMERIC,
|
|
|
|
|
ACTUAL_VALUE_STORED VARCHAR,
|
|
|
|
|
NEW_EXTRACTED_VALUE VARCHAR,
|
|
|
|
|
CONFIDENCE_LEVEL NUMERIC,
|
|
|
|
|
SNIPPET VARCHAR,
|
|
|
|
|
ORIGINAL_PAGE_NUMBER VARCHAR,
|
|
|
|
|
NEW_PAGE_NUMBER VARCHAR,
|
|
|
|
|
REVISED_PROMPT VARCHAR,
|
|
|
|
|
RESULT BOOLEAN,
|
|
|
|
|
LOAD_DT TIMESTAMP
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
-- Create file format
|
2024-03-15 16:31:43 -05:00
|
|
|
-- CREATE FILE FORMAT IF NOT EXISTS STG.CSV_HEADER
|
|
|
|
|
-- TYPE = 'CSV'
|
|
|
|
|
-- FIELD_DELIMITER = ','
|
|
|
|
|
-- FILE_EXTENSION = '.csv'
|
|
|
|
|
-- RECORD_DELIMITER = '\\n'
|
|
|
|
|
-- DATE_FORMAT = AUTO
|
|
|
|
|
-- TRIM_SPACE = TRUE
|
|
|
|
|
-- NULL_IF = ('NULL', '', 'N/A','?','~','\\N')
|
|
|
|
|
-- SKIP_HEADER = 0
|
|
|
|
|
-- EMPTY_FIELD_AS_NULL = TRUE
|
|
|
|
|
-- FIELD_OPTIONALLY_ENCLOSED_BY = '"'
|
|
|
|
|
-- SKIP_BLANK_LINES = TRUE;
|
2024-02-27 14:29:22 -06:00
|
|
|
|
|
|
|
|
|
|
|
|
|
-- Create staging table for attempt logs
|
|
|
|
|
CREATE TABLE IF NOT EXISTS STG.TRAINING_ATTEMPT_LOGS
|
|
|
|
|
(
|
|
|
|
|
FIELD_NAME VARCHAR,
|
|
|
|
|
CONTRACTS_TESTED NUMERIC,
|
|
|
|
|
USERNAME VARCHAR,
|
|
|
|
|
DATE_TIME TIMESTAMP,
|
|
|
|
|
ACCURACY NUMERIC,
|
|
|
|
|
ATTEMPT_NUM NUMERIC,
|
|
|
|
|
LOAD_DT TIMESTAMP
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
|
2024-03-14 14:23:20 -05:00
|
|
|
|
|
|
|
|
-- Create table for column config that will be used by SP to create raw training data table
|
|
|
|
|
CREATE TABLE IF NOT EXISTS STG.TRAINING_DATA_COLUMN_CONFIG(
|
|
|
|
|
COLUMN_NAME VARCHAR,
|
|
|
|
|
COLUMN_DATATYPE VARCHAR
|
|
|
|
|
);
|
|
|
|
|
-- Raw training data table is created dynamically with the help of stored procedure
|