I read Data Points by Nathan Yau this weekend – fantastic read.
All you really need to know is that I couldn’t put the book down. It’s so easy to read and he’s picked amazing examples to illustrate an “Introduction to Data Visualization”.
Here are some points which stood out to me:
List of questions to ask when exploring a dataset
At the Data School, we see a lot of amazing vizzes. It’s intimidating when we see a lot of intelligently designed and thoughtfully put together pieces and be expected to put together such things. Then when you’re given a massive dataset, you’re just swimming in numbers. There’s a section dedicated to “How to Explore a Dataset” which has a fantastic chart of the possible questions, its related statistical concepts and possible visualizations. Kind of wish this was the Data School textbook or at least a necessary pre-read for the next class.
Great sources for datasets
I’m always on the lookout for good quality, robust datasets. I’ve noted to look up the Centre of Disease Control has as well as the US Census Data‘s collection of shape files (The UK’s ONS also has this as well).
Understanding a heatmap as a new way to visualize scatterplots
As someone who love’s scatterplots, the question then becomes how to “…so how do I plot more measures?”. The parallel coordinates plot in the form of a heatmap is something I think I’ll try to recreate in Tableau.
Visualizations without numbers
Nathan brings up an interesting point about how aggregations (averages for example) can cover up telling data. Sometimes it is necessary to introduce many many marks to the visualization to order to not let the aggregation skew important granular patterns. I think this may be an interesting idea to explore for high level KPI dashboards in industries with many different fast moving pieces of data like airlines to see patterns (i.e. seeing many departure dates instead of aggregating to an average for each flight). I’ll be trying this out too when I find a bit of time.