Read a CSV from standard input and classify every column:
- numeric column:
keyif its name ends withid(case-insensitive) and its values are unique, otherwisequantitative; - text column:
temporalif every non-missing value parses as a date-time (pd.to_datetime(..., errors="coerce", format="mixed")); otherwisekeyif the name ends withidand values are unique; otherwiseordinalif all lower-cased values belong to {low, medium, high, good, moderate, unhealthy, small, large}; otherwisecategorical.
Print 4 items x 5 attributes, then per column pm25: quantitative (Q), 3 distinct, 1 missing where the code is N for key and categorical, O, Q or T.