Dataset collection

Dirty Datasets for Data Cleaning Practice

Practice finding duplicates, missing values, inconsistent labels, and other data issues with documented challenges for SQL, Excel, and Python.

Compare the available challenges by issue type and the relationships involved. Start with a small, clearly defined problem before combining several cleaning steps.

Compare your options

Choose a dataset for the work you want to complete

Dataset	Documented problems	Dirty variant	Reference
Ecommerce Sales & Returns	Duplicates, inconsistent labels, date formats, and blank required values	Included	Unchanged generated release
SaaS Subscriptions & Churn	Duplicates, inconsistent labels, date formats, and blank required values	Included	Unchanged generated release
Marketing Campaigns & Conversions	Duplicates, inconsistent labels, date formats, and blank required values	Included	Unchanged generated release
Inventory & Supply Chain	Duplicates, inconsistent labels, date formats, and blank required values	Included	Unchanged generated release
Workforce & Employee Attrition	Duplicates, inconsistent labels, date formats, and blank required values	Included	Unchanged generated release

Start here

Challenge issue checklist

Each published synthetic cleaning challenge documents the same issue classes. Find the affected rows rather than relying on a row-by-row answer key.

At least one duplicated source record in each table
Casing or whitespace inconsistencies in selected descriptive fields
Slash-delimited dates in selected date fields
Blank values in selected required non-key fields
Before-and-after row counts and totals for every correction

Practical guidance

Start with a useful result

Keep an explanation of every change

Preserve the original file, identify the problem, and state the rule used to correct it. Check whether the change affects row counts, totals, or relationships. Your final result should explain both what changed and why.

Unusual does not automatically mean incorrect

A negative amount may represent a refund. Repeated values may belong to separate transactions. Missing information may be meaningful. Use the field definitions and challenge notes before replacing values or removing rows.

What is the difference between a source file and a cleaning challenge?

Source files represent the published data before your own analysis. A cleaning challenge is a separate practice version with intentional problems. Use its notes and reference data to understand what you are expected to investigate.

Learn to explain the fix

Go beyond spotting an error. Practice the transformations and checks that show a correction did what you intended.

Explore data-cleaning practice