Garbage in, garbage out
Even the best AI will not give accurate analysis from a chaotic table. Before calculating and analyzing, the data needs to be put in order. AI helps with the cleaning itself, but first let us understand what "clean data" is.
Signs of a tidy table
- Column headers: the first row is clear names.
- One type per column: "Date" has only dates, "Amount" only numbers.
- A single format: dates the same way, numbers with no extra symbols.
- No empty rows/columns in the middle of the data.
- One cell, one value: not "Ivan, Moscow, 30" in one cell.
AI helps clean
In the table dates are written differently: "01.02.2025", "1 Feb", "2025-02-01". How do I bring them to one format?
A common trouble. Options: 1) select the column to Format Cells to Date, one template (if the system recognizes it). 2) If it does not recognize ā use a function or replace to bring to one format. 3) AI helps: describe which formats exist and which you need, and I will suggest a normalization formula. The main rule for the future: enter dates in one format from the start.
Typical problems and how to spot them
- Numbers as text: they do not sum ā often due to spaces or commas.
- Duplicate rows: distort counts ā AI helps find and remove them.
- Extra spaces: "Moscow" and "Moscow " count as different.
Describe the problem to AI ("numbers do not sum", "there are duplicates") ā it suggests the cause and a fix. Cleaning at the start saves hours of troubleshooting later.
Rule: 80% of analysis errors come from dirty data, not formulas. Order the table first, then calculate.
š§ Why does data cleanliness matter before analysis?