Skip to main content

issues()

List the issues that the conformance check found.

Usage

Source

issues(
    severity=None,
    status=None,
)

An issue is a rule (or, for a Validate-based report, a validation step) that found something wrong. The list is a quick way to see what has to be fixed, and it can be turned into a DataFrame or saved. Each issue is a dict, whose keys depend on the form of the report.

Parameters

severity: str | None = None

For a Validate-based report, only keep the issues of this severity: "warning", "error" or "critical". The severity of a step comes from the thresholds= of validate_conformance(), so without thresholds no issue has a severity. This is ignored for the other forms of report. The default is None, which keeps all the issues.

status: str | None = None
For a CDISC CORE report, only keep the issues of rules with this status, such as "ISSUE REPORTED". This is ignored for the other forms of report. The default is None, which keeps all the issues.

Returns

list[dict]

For a report of the built-in rules, a dict for each rule with issues, with the keys dataset, rule_id, rule_type, message, n_issues, sensitivity ("Error" or "Warning") and status.

For a CDISC CORE report, a dict for each rule and dataset with issues, with the keys dataset, rule_id, message, issues (the number of issues) and status.

For a Validate-based report, a dict for each step with failing test units, with the keys dataset, step, assertion, column, n_failed, n and severity.

Examples

Let’s check a small DM dataset with the built-in rules and list the issues as a DataFrame:

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})

pl.DataFrame(report.issues()).select("rule_id", "dataset", "sensitivity", "n_issues")
shape: (10, 4)
rule_iddatasetsensitivityn_issues
strstrstri64
"SDTM-007""DM""Error"1
"SDTM-011""DM""Error"1
"SDTM-031""DM""Error"1
"SDTM-032""DM""Error"1
"SDTM-033""DM""Error"1
"SDTM-034""DM""Error"1
"SDTM-035""DM""Error"1
"SDTM-121""TS""Error"1
"SDTM-122""TA""Warning"1
"SDTM-195""DM""Error"1

The missing TS and TA datasets are given as the datasets of SDTM-121 and SDTM-122. Only SDTM-122 is of "Warning" sensitivity. A Validate-based report gives the severity of each step instead, when thresholds are set:

ae = pl.DataFrame({"USUBJID": ["STUDY01-001", "STUDY01-009"], "AESEQ": [1, 1]})
study = pb.SubmissionPackage(datasets={"DM": dm, "AE": ae})

report = study.validate_conformance(engine="validate", thresholds=pb.Thresholds(warning=1))

report.issues(severity="warning")
[{'dataset': 'AE',
  'step': 5,
  'assertion': 'specially',
  'column': None,
  'n_failed': 1,
  'n': 2,
  'severity': 'warning'},
 {'dataset': 'DM',
  'step': 10,
  'assertion': 'col_vals_expr',
  'column': None,
  'n_failed': 1,
  'n': 3,
  'severity': 'warning'},
 {'dataset': 'DM',
  'step': 11,
  'assertion': 'col_vals_regex',
  'column': 'RFSTDTC',
  'n_failed': 1,
  'n': 3,
  'severity': 'warning'}]

The issues are the orphan subject in AE, and the length of "FEMALE" and the date "15JAN2024" in DM. Each of them is one failing test unit, which reaches the warning threshold.