Quantitative data uses numerical values to check whether a dataset is complete, plausible, and useful before analysis. Start with record counts, then inspect samples, fields, years, and missing or suspicious values. section 5.2.2
Use a short validation checklist:
- Count records and compare the result with the expected time span, reporting frequency, and sensor population. section 5.2.2
- Inspect a handful of rows, probe different years, verify the schema, and investigate unexplained fields. section 5.2.2
- Measure null and zero temperatures separately rather than assuming every numerical value is valid. section 5.2.2
Smart-building sensor readings
A five-year table contains 262,944,000 records, consistent with 1,826 days and 144,000 readings per day. A sample row includes a temperature and an unexplained “op” field, which becomes a backlog investigation item. section 5.2.2
select count(*) from temperature_readings >262944000
Use the count query as the first compact check of dataset scale. section 5.2.2
