Module 05 Summary

What this module established

The first question about a missing value is why it is missing, not how to fill it. A form never submitted and a ward that was closed both show as NA and mean opposite things; filling both with zero asserts one of them for both.

Carry forward

• A duplicate is a repeated key, not a repeated row. duplicated() finds only the perfectly identical ones, which are rarely the ones inflating your totals.
• Keep validation as a check column so failures can be counted. Impossible values a rule can handle; implausible-but-legal ones need a human.
• Category columns split silently on whitespace and case. Run count() before grouping and confirm the number of categories.
• Declare factor levels for ordered categories and state the format when parsing dates — 01/03/2026 is two different days.

Before moving on

Move on when your cleaning log accounts for every row that left the table.

Mark complete