Get the failing records that the conformance check found.
A rule that checks the values of a dataset (such as a codelist check on SEX) fails for particular records, and each of these is a finding. The findings() method gets them all as a list of objects. For the built-in rules, findings_df() gives the same findings as a DataFrame, which is often easier to work with.
Returns
list
-
For a report of the built-in rules, a NativeRowFinding for each failing record, with the attributes rule_id, dataset, row (the 0-based position of the record), usubjid, checked_column, checked_value, context (the record’s identifying variables, such as STUDYID) and message. At most 100 findings are kept for each rule.
For a CDISC CORE report, a CoreFinding for each entry of CORE’s Issue_Details, with the attributes rule_id, message, dataset, executability, usubjid, row (the 1-based position of the record), seq, variables and values.
For a Validate-based report, an empty list.
Examples
Let’s check a small DM dataset with the built-in rules:
import pointblank as pb
import polars as pl
dm = pl.DataFrame(
{
"STUDYID": ["STUDY01"] * 3,
"DOMAIN": ["DM"] * 3,
"USUBJID": ["STUDY01-001", "STUDY01-002", "STUDY01-003"],
"RFSTDTC": ["2024-01-15", "15JAN2024", "2024-01-17"],
"AGE": [34, 51, -38],
"SEX": ["M", "F", "FEMALE"],
}
)
report = pb.validate_sdtmig({"DM": dm})
for finding in report.findings():
print(finding.rule_id, finding.row, finding.usubjid, finding.message)
SDTM-007 2 STUDY01-003 SEX value is not in the SEX codelist.
SDTM-011 1 STUDY01-002 RFSTDTC does not conform to ISO 8601 extended datetime format.
SDTM-195 2 STUDY01-003 AGE must not be negative in DM.
Ten rules fail, but only these three check values. The others fail because a variable or a dataset is missing, so they have no failing records.