Event types
5
Shared columns, different meanings
Executive summary
What this dataset says
37.1% of source cells are blank, but 24 blanks occur in fields that are always empty within their event type. Review the schema before treating them as data loss.
Days with records
2 / 2
Presence does not prove full-day coverage
Type-pattern blanks
24
26 total blank cells
Cross-midnight sleeps
1
Do not assign whole duration to start date
Evidence
Findings supported by the data
0124 / 26
Validate completeness by event type
A blank end time on a feed is not equivalent to a blank end time on a sleep. Always-empty fields suggest conditional applicability, not proof that they are optional. Confirm the source schema; this small sample cannot establish required fields.
02Date-boundary audit
Midnight boundaries can distort daily totals
1 sleep records cross calendar dates. Split intervals at midnight before comparing calendar-day sleep. Check time zone, overlapping intervals, and units before summing durations.
03Logging coverage is unknown
Do not confuse an unlogged day with zero activity
0 dates inside the observed span have no records. Even active dates may be partial. No daily baseline, health interpretation, or feeding/sleep relationship is claimed from incomplete logging.
Composition
Recorded events by type
Action plan
Recommended next steps
Review the evidence before acting
Use the calculations, scenarios, and limitations in the executive summary. Validate denominators and row grain before making decisions.
Method & scope
How to use this report
Analysis performed
Patlr profiled 10 rows and 7 source fields, detected 5 record groups, and added 8 analysis-ready fields while preserving every original column.
Interpretation boundary
Results describe patterns in recorded events. They do not establish causation and are not medical guidance. Missing or inconsistently recorded events can affect daily measures.
Evidence-only mode
Patlr does not add external facts. AI-assisted classifications are retained only when paired with an exact quote from the submitted text; unsupported output is left blank. Interpretations should still be validated before decisions.