I’m really tired of line graphs, pie charts, and other analytical eye candy. I can’t do anything with the questions they present. It is always impossible to drill into the data that creates these confusing images.
I have been analyzing complex data for years. I have have crunched data to understand everything from signals intelligence to sales performance. Ironically, the most useful methodology always boils down to three distinct concepts that work beautifully together.
1. Lists-Discrete elements neatly listed. These could be URLs, Twitter users, contacts, songs, keywords, or any of an infinite number of data items. By breaking data into its discrete element we can see patterns and trends more easily. Simple, easy to scan and sort, lists become agile frameworks to manipulate and analyze.
2. Counts-Frequency, scale, reputation are all potential revelations from the simplicity of counting occurrences. Counts make it easy to reveal what is most relevant in data sets. What’s more counts create relevant lists of associated elements.
3. Nodes-Where is the data coming from, going to, or inter-related? This is a powerful way to gauge the quality and accuracy of information. There is a tendency for associates to exchange data and that exchange tends to be consistent. Therefore, following nodes and relationships can help you locate the right data.
These three data structures give you all the elements quickly and powerfully analyze very large data sets. Next time you are building an analytics dashboard start here and don’t necessarily go beyond.
Everything else is pretty, but probably far less useful and ultimately frustrating.