Volume 1: Lessons from the Trenches of Applied Data Science
Buy Volume 1
Get the ebook from LeanPub (preferred)
The ebook is available in multiple formats, DRM-free, for “pay what you can” with a suggested price of $15
Also available now in Kindle and in print through Amazon.
Suggested Citation
Geller, Wendy, Cratty, Dorothyjean, and Knowles, Jared. 2019. Education Data Done Right: Lessons from the Trenches of Applied Data Science. Victoria. Leanpub. Available online: www.eddatadoneright.com
Chapter Summaries
Metadata
Following the Introduction, in Chapter 2, we begin where all analysis should—with the metadata. Metadata is how we know what our data mean and how we can evaluate their provenance.
Analyst’s Guide to IT
After covering metadata in all its glory, we move on to talk about how analysts and IT departments can work together to improve both the metadata and the operating efficiency of analysts in Chapter 3.
Data Requests
In Chapter 4, we switch gears and start talking about practical tasks as a way of illustrating how metadata and IT operations play out in the day-to-day work of the agency. Data requests for research and analysis provide an excellent case study for how agencies use data. Those requests always touch on issues of metadata, data governance, and IT operations
Bureaucracy and Politics
We then turn our attention to navigating the organization itself in Chapter 5, Bureaucracy. In this chapter we cover some of the main concepts that help in understanding why an organization operates the way it does, and what we as analysts can do within that organization.
The Power of Descriptives
Finally, in Chapters 6 and 7 we talk about the power of descriptive analyses in an agency. While an awful lot of the focus in our training tends to be on inferential methods, much of the demand from within our organizations is for high quality descriptive work. Both chapters include some concrete examples and illustrations, and the second chapter takes the conversation a bit further, by providing detailed statistical program command lines for dozens of specific descriptive tools
Table of Contents
1 Welcome
1.1 Introducing Education Data Done Right
1.2 So, who are we and why did we write this book?
1.3 Why you should keep reading this book
1.4 What’s in this first volume, anyway?
1.5 What we hope you’ll take away from this book
2 The Holy Grail of Data Science: Rock Solid Metadata and Business Rule Documentation
2.1 Introduction
2.2 Metadata and You (this is a forever relationship)
2.3 The Living Data Dictionary: Syllabus to Your Work
2.3.1 What Does a Good One Look Like?
2.4 Data Management
2.4.1 Sound Analysis: Grinding for the Good of the Metadata
2.4.2 Provide Continuity
2.4.3 How You Can Do It
2.4.4 Tools to Consider
2.5 Documented Business Rules: The Data Scientist’s Handbook
2.6 Conclusions
2.7 Appendix
3 An Analyst’s Guide to IT
3.1 Introduction
3.2 Groundwork
3.3 IT 101: Speaking the Same Language
3.3.1 Business Rules
3.3.2 Change Management
3.3.3 Database
3.3.4 Data Governance
3.3.5 Data Integrity
3.3.6 Data Warehouse and/or Operational Data Store
3.3.7 Enterprise
3.3.8 ETL
3.3.9 Version Control
3.4 IT 102: Strategies for Successful Collaboration with IT
3.4.1 Bring IT in From the Beginning
3.4.2 Play Up the Cool Factor
3.4.3 Get Creative with Resources
3.4.4 Use Their Processes and Procedures to Get it Right
3.4.5 Build Informal and Formal Channels of Communication
3.5 Conclusion - What to Take Away From This Chapter
4 Data Requests: You Can Make Them Useful (we swear)
4.1 Introduction
4.2 Taking a different perspective
4.3 Prioritizing and Triaging Requests: Data Governance to the Rescue
4.4 Check Before You Wreck: Why Levels of Granularity Matter
4.5 How to Respond to the Request
4.6 Managing Data Requests
4.6.1 Publish Data
4.6.2 Collect Requests in a Single Place
4.6.3 Inventory Common Requests
4.6.4 Automate Common Requests
4.6.5 Use Request Approval to Build Data Governance Decisions
4.7 Use the Data Sharing Agreement for Good
4.8 Closing Thoughts
5 Politics and Data Driven Decision Making
5.1 Introduction
5.2 What Do We Mean By “Politics”?
5.3 Politics 101
5.3.1 Policy Windows
5.3.2 Credit Claiming and Failure Blaming
5.3.3 Loss Aversion and Incrementalism
5.3.4 The Role of Information
5.4 Key Practices
5.4.1 Humility
5.4.2 Repeated Engagement
5.4.3 Coalition Building
5.4.4 Reputation Management
5.4.5 Timing
5.5 Conclusion
6 Moments of Truth: Descriptive Statistics are Some of the Most Important Work You’ll Do
6.1 Introduction
6.2 Why write this chapter?
6.2.1 Because the data’s origin story matters
6.2.2 Because modeling is better with fewer guesses
6.3 The power of descriptives
6.4 Describing your data early and often
6.5 Three important data stages
6.6 Three handy rules for leveraging descriptives
6.7 Conclusion - Words can’t describe
7 Applying Tools of the Trade: Descriptive Data Commands in Context
7.1 Introduction
7.2 Describing your data – what this looks like in practice
7.2.1 Two types of descriptives
7.2.2 Data components or elements: variables, values, and observations
7.2.3 Command-relevant shorthand for data elements
7.2.4 Descriptive tools
7.2.5 Descriptions: summary statistics (frequencies, crosstabs, tables of means)
7.2.6 Depictions: graphs of distributions and relationships (histograms, bar graphs)
7.3 Conclusion
8 Conclusion – Summaries of Chapter Lessons Learned
8.1 Metadata
8.2 Analyst’s Guide to IT
8.3 Data Requests
8.4 Bureaucracy and Politics
8.5 The Power of Descriptives