patterns, analysis, and metrics

I absolutely love finding patterns and causes; while distinct entities, they are what make analysis interesting, and what everyone who "hates research" tends to overlook. This set of tea canisters, for example. I spent a good amount of time thinking through the patterns that make up the spatial positions throughout the wall, qualitfying and quantifying them the best I could. It was almost better than the tea itself.

The patterns and canisters point to the most fascinating part about data analysis: nothing is random; there is only cause and effect. Everything is human, from the arrangement of physical objects to the arrangement of our perceptions, beliefs, and preferences. As an example, try finding the underlying causes of the nature of data collected on Daytum. Daily intake of food and drink is a good start.

The while pattern-finding is fascinating, the hardest part of data analysis is never the analysis itself. It's always determining the metrics. Far more often than not, the metrics we choose often assume the cause, as opposed to reaching to get a critical understanding of the group in question. Consider the idea of determining the reason people like a particular soda by measuring the amount of soda a group of people drink over time. Or by qualifying the reasons they reportedly like soda. The problem is that the measurement assumesthat the reason people drink a particular soda is because of its characteristics. Certainly this is a rational approach but since when have our desires ever been rational?

You could measure when they drink the soda. After what experiences? When did they first start? You'd be getting on the right track, but almost always we devise metrics by subconsciously asking ourselves, "if X is the reason people drink this soda, how do I measure X?" Begging the question at its finest.

It's a daunting task, but it's only when we gather all the information - not just the information we think is important - that the real interesting patterns start to emerge. How else would we know that road closures actually lead to fewer traffic jams? (much more important: why)

Sounds suspiciously like "observe everything." The general theme here: human behavior is of course far too counterintuitive for simple, top-of-mind metrics. 

So why all this 3rd-grade level ROI out there?

Comments

[Olle]

Yup, that's interesting and true. I just spent a couple of days in Copenhagen on vacation. And my vacations tend to be walking around with a camera, sitting in bars and cafes, just watching. Human behavior really is counterintuitive and impossible to simplify (at first). I like the theme observe everything. Most important for a planner. Always on in order to find the interestingness between the specific and the general.