Invalid values in the data

Look at this table of employees:

name	age	height (cm)	favourite colour
A	11	145	blue
B	200	150	green
C	12	1.48	red
D	11	152

Three problems: B is 200 years old, C is 1.48 cm tall, D has no colour.

• Impossible values (200 years old) — someone made a typo.
• Wrong unit (1.48 is metres, not cm).
• Empty values — the question was skipped.

How errors happen: carelessness during data entry, wrong units, a broken sensor, or an unclear question.

What to do? Never just delete silently. Choose and write down what you did:

• correct it if you know the right value (1.48 m → 148 cm),
• remove the row if the value is unusable,
• or leave it empty and calculate without it.

An algorithm trained on the table assumes that everything in it is true. If you write 200 years, the model learns that 200-year-olds exist. Garbage in, garbage out.