Data Cleaning and Validation for Analysts cookbook

A new export lands in your inbox and everyone wants the report today, with no way to know yet if a single number in it is real.

A column is missing a chunk of its values and three people on the team each fill the gaps a different way.

You cleaned a file three weeks ago, someone asks how you fixed the missing region codes, and you genuinely cannot remember.

Every column passes the obvious checks and the numbers are still wrong, in a way nobody notices until someone charts the values.

One file writes dates as day-month-year and prices in euros, the other writes month-day-year and dollars, and nothing lines up once you stack them.

A report's grand total and the sum of its own rows have drifted apart, and nobody would notice until someone does the math by hand.

One row sits ten times bigger than everything around it, and deleting it feels risky either way.

The dashboard number looks reasonable until someone points out it's exactly double what finance has.

Finance's number and your dashboard's number are both confident, and they disagree by four figures.

The latest customer satisfaction survey looks great, and only the people who never canceled ever got a chance to see it.

Marketing swears the customer list is clean, and 'Jon Smith' at one address sits three rows away from 'John Smith' at almost the same one.