Writing

Jon Reades & Fulvio Lopane

“I used to think that writing is the most trivial part of the job of a data scientist; because the analyses and numbers should speak for themselves right? Wrong!”

This is from an article (see references at the end) by a practicing data scientist on the importance of communication skills. Their argument, with which I’d broadly agree, is that:

1. Writing is used everywhere the corporate world.
2. Writing brings clarity to your thinking process.
3. Writing is the ‘last mile’.

I think there’s a lot you can take away from that article, so I’d strongly encourage you to read it for yourself. Notice the focus on the ‘so what’ question and the emphasis on story telling. If you can you do this now, you’ll find the dissertation that much easier.

Keep it short. Keep it simple. Keep it direct. Signpost. Don’t try to use complicated vocabulary where plain-English will do. This is something that ChatGPT gets wrong a lot when you ask it to do ‘academic writing’.

Use figures where they replace a paragraph (or more) of content. Use a sentence where it can replace a figure (or more) of content.

What?

(Does a Spatial Data Scientist do?)

Ask students what they think a data scientists does.

Good answers can be added below.

The ‘Pipeline’

• Code
• Analyse
• Understand
• Communicate
• Report

Here are five things that we might imagine any data scientists, spatial or otherwise, does.

Do we think they’re in the right order?

I can tell you right now that this sequence of steps is how to get a mark of between 45 and 63 in your dissertation.

The ‘Pipeline’

• Understand
• Code
• Report
• Analyse
• Communicate

Here’s a slightly better order, here’s why:

• You can’t answer a question – from your boss, from an academic, from your friends even – if you don’t understand it. So before you run off and start writing some code, the first thing you need to do is understand the problem you’re trying to solve. That problem should not be technical, it should be practical.
• Once you understand the problem you can start trying to code a solution.
• The code will allow you to produce reports. These reports might be to do with data quality, they might be diagnostics from a Random Forest Machine Learning algorithm. At each stage in the development of your results you should be generating reports that help you to better-understand your problem and work out if your code is working.
• Once the reports have given you confidence in your findings now you can actually write the analysis. Your analysis might lead you to realise that you need to go back and write more code and produce more reports, but that’s normal.
• Finally, you need to work out how to communicate your analysis. If you understand the problem then you’ll find this process rewarding. If you don’t then you’ll find it frustrating and want to brush it off quickly.

So the fact these are in a list is still rather misleading because at each point you get feedback effects, and it’s also a loop.

Another Way…

Another way to think about all this is to write backwards:

How to Understand?

How should we do this?

How to Understand?

How should we do this?

Why are we doing this?

How to Report?

This is the number.

How to Report?

This is the number.

These are the takeaways.

How to Analyse?

These are the methods we can use.

How to Analyse?

These are the methods I can use.

This is the method that matches the need.

How to Communicate?

What do I need to say?

How to Communicate?

What do I need to say?

Who needs to know?

On Writing

Writing == Thinking

If writing down your ideas always makes them more precise and more complete, then no one who hasn’t written about a topic has fully formed ideas about it. And someone who never writes has no fully formed ideas about anything nontrivial. ~ Paul Graham

Writing is thinking. To write well is to think clearly. That’s why it’s so hard. ~ David McCullough

Clarity == Understanding

If you can’t explain an idea to an 8 year old, you don’t understand it. ~ Richard Feynman

I Don’t Care

Focus on the ‘so what’.

Tell Me a Story

Don’t bury the lede.

All data science is, ultimately, a story. A story of struggle. Frustration. Discovery. Learning. But you need to tell that story the right way.

Many of you will have learned some ‘system’ for writing in school. The inverted pyramid or something like that. In university, in my literary theory class I picked up the pyramid approach: taking a single sentence and unpacking that into the themes of the entire book.

There are mystery novels. Romance novels. Economist articles. Teen Vogue articles. They are all telling stories. They all do this in different ways.

Tell the Right Story

(That’s why we’re torturing you.)

So what’s fine for consultancy, might not be for academia or newspaper reporting. And vice versa.

Across the last five years what we’ve noticed

Resources

• The Most Undervalued Skill for Data Scientists
• One Mindset Shift That Will Make You a Better Data Scientist
• Sword, H. (2017). Air & light & time & space: How successful academics write. Harvard University Press.¹
• Greene, A. E. (2013). Writing science in plain English. University of Chicago Press.²

1. Available for free via JStor.

2. Not seemingly available for free, but I found a nice little summary (with typos) here.