How I Measure Data Quality (And How You Can Too)

Hi everyone! I'm a senior data scientist with over ten years in the field, and there's one question that pops up all the time: "How do I know my data quality is good enough?" If you've struggled with that too, this post is for you.

Metrics, Metrics, Metrics

The key is to measure your data on those quality dimensions that really matter for what you're doing. You need numbers – that's the only way to get a reliable picture. So, I highly recommend using data quality KPIs (Key Performance Indicators).

Data Quality: It's All About Purpose

Data is good enough when it works the way it's supposed to. Picture this: You've got a fancy real-time fraud detection system. Speed is everything, right? You need super fresh data. So for you, 'timeliness' is your most important data quality metric.

But what if your focus is keeping customers happy? Now accurate information is crucial. You can't send offers for products that aren't in stock just because your tables are mismatched. In this case, accuracy and validity steal the spotlight.

Introducing Data Quality Dimensions

Here's where things get organized. Think of these as categories of common problems:

• Timeliness: Is your data up-to-date?
• Validity: Are values in the right format?
• Accuracy: Does the data reflect reality?
• Completeness: Any missing info?
• Uniqueness: Are there annoying duplicates?

KPIs: Your Data Quality Scorecard

Here's how to actually measure data quality:

1. Set up Checks: What do you need to monitor?
2. Match to Dimensions: Which dimensions do your checks fall under?
3. Count the Wins: Track how often your checks pass.
4. The Percentage Power: That's your KPI – percentage of passed checks (over a day, week, etc.)

Each dimension can have its own KPI. Say you get these results:

• Timeliness: 90% (Yikes, missed some opportunities!)
• Validity: 95% (Still, 5% of customers getting bad info isn't cool)

Below are the top 10 online references that can provide valuable insights into creating and implementing data quality checklists to ensure reliable analytics. These sources include a mix of articles, industry guidelines, and academic papers that are highly regarded in the field of data science:

Top 10 References on Data Quality Checklists for Reliable Insights:

1. "Data Quality: The Accuracy Dimension" - Jack E. Olson

   • This book provides an in-depth look at data quality with a focus on accuracy, including practical frameworks for assessing and improving data quality.
   • Link to the book

2. "Data Quality Assessment" - Arkady Maydanchik

   • Arkady Maydanchik offers methodologies for data quality assessment that are crucial for any organization looking to ensure the reliability of their datasets.
   • Link to the book

3. "Executing Data Quality Projects: Ten Steps to Quality Data and Trusted Information™" - Danette McGilvray

   • This book outlines a ten-step process for planning and implementing data quality projects, which can be extremely useful for creating a checklist for data quality.
   • Link to the book

4. "The DAMA Guide to the Data Management Body of Knowledge (DAMA-DMBOK)" - DAMA International

   • This comprehensive guide covers various aspects of data management and includes sections on data quality management.
   • Link to purchase or access

5. "Juran's Quality Handbook: The Complete Guide to Performance Excellence, Seventh Edition" - Joseph A. Defeo

   • Although not specifically about data quality, this handbook contains essential principles of quality management that can be adapted to data quality initiatives.
   • Link to the book

6. "Measuring Data Quality for Ongoing Improvement: A Data Quality Assessment Framework" - Laura Sebastian-Coleman

   • This book introduces a framework for assessing the quality of data, which is crucial in developing effective data quality checklists.
   • Link to the book

7. "Improving Data Warehouse and Business Information Quality: Methods for Reducing Costs and Increasing Profits" - Larry P. English

   • Offers insights into improving the quality of data in data warehouses and business intelligence systems.
   • Link to the book

8. "Critical Data Studies: An Introduction to the Critical Role of Data in Society" - Edited by Dalton and Thatcher

   • This collection of essays provides critical insights into the implications of data quality in societal contexts.
   • Link to more information

9. "Data Quality for Analytics Using SAS" - Gerhard Svolba

   • This book is specific to using SAS for data analytics but provides general principles on ensuring data quality that can be applied broadly.
   • Link to the book

10. "Practical Data Migration" - Johny Morris

    • Provides a practical approach to data migration, which includes crucial steps for data quality checking.
    • Link to the book

These resources offer a broad perspective on data quality, from theoretical frameworks and methodologies to practical tips and industry-specific guidelines.

Did I Forget Anything?

That's the basics of measuring data quality! Let me know if you have any other tips or favorite techniques.

Let's get those data problems sorted!
Best,
Kemal Cholovich

dataquality #datagovernance #dataengineering #datascience