Keeping up with data — Week 2 reading list
5 minutes for 5 hours’ worth of reading
My week, apart from being extremely busy, was epitomised by data visualisations. Until this week, I’d found most of the visualisations I enjoyed either through The Big Picture, FiveThirtyEight or randomly. Whereas, I stumbled across the visualisation of 645 hours of watching series! It is noteworthy for two reasons. Firstly, it reminded me how lucky I have been with Covid-restrictions in Switzerland having spent almost the same time on my bike last year. But more importantly, it navigated me to new sources of great data visualisations — @VeryData_365, #MakeoverMonday and the whole community around it.
Other topics on this week’s reading list include organisation of data teams and specifics of data science projects. Enjoy!
- Cultivating algorithms: Stitch Fix believes in the value data science. But they are also aware of its specific needs for iteration, exploration, and learn-as-you-go development driven by curious teams. That has led them to a unique design of the organisational structure, roles and processes of data science operations. It is written as one of these popular scrollable stories, which I personally find so difficult to read. But if you won’t get distracted by objects flying all over your screen, you’ll learn a lot of insights about setting up and operating data science, or how an organisation can learn to embrace uncertainty. (Stitch Fix)
- Organizing Data Teams — Where to Make The Cut: How to structure data teams is a conundrum faced by many larger organisations. Should they be centralised? Decentralised? Or is a hybrid combination the best solution? Since context matters a lot, there is no right answer. When reading the article, I realised that next to the standard trade-off — domain knowledge vs. data science expertise — the ‘status of data’, i.e., how important is data for the organisation, and the level of data-maturity play a huge role. In the beginning you might want to build a centralised department to quickly prove the value of data. Then, decentralise it in time to spread the benefits within the organisation. But be ready to centralise back if it turns out as necessary to speed things up. (Sven Balnojan @ Towards Data Science)
- Why Data Science Development Process is like Playing Game Boy® Final Fantasy: Data science projects are notoriously difficult to manage because they deal with highly non-linear, heavily iterative, ‘R&D’-type process. Understanding and acknowledging this is a great first step in setting yourself up for success. Potentially, not everyone will enjoy the Final Fantasy Legend II analogy used in the article, but I certainly did. As a data professional I know that I need to have a whole arsenal of analogies and examples to demonstrate the complexity of data science project management. This is not a topic you can escape. (Bill Schmarzo)
- How to start building a new dashboard for clients? After running into the image above I clicked through to the related blog post and also went through a couple of older articles. The one that caught my attention was a guide or checklist to be used when building a new dashboard for clients. I really appreciate when experts put these checklists together. When reading them, it often feels too obvious. But the added value lies in the fact that, as easy as it is, omitting just one step could lead to painful consequences. And the fact that an expert has put the points on the list means that she has already learned these lessons and is forewarning us of falling into the same traps. (Data muggle doing magic)
- 📊 I Studied 365 Data Visualizations in 2020: What can be learnt from analysing a visualisation every day? You can gather high-level trends — people liking maps, dark mode, animated visualisations or polar maps. But you can also discover new types of visualisations, such as words, scrollable stories or augmented reality data visualisations. Since data visualisations are everywhere, it is good to keep an eye on the latest trends. (Avery Smith @ Towards Data Science)
This week I also saw a lot of armchairs in the shape of an avocado, which was understandably trending heavily! Thankfully, I saw less of the LinkedIn ‘surveys’ about preferences for working from home / office / both, years when we started working or if we prefer Mac to Windows. Given my obsession with using data to solve problems I can think of only a couple of obscure reasons why people should want to know the answers to these questions.