Thoughts on Activity Heatmaps

So, I've been really itching to do some kind of analysis of what my best times of day are (the article that inspired this desire: How Heatmapping Your Productivity Can Make You More Productive), and so far all I know is that I have a LOT of data (yay, BubbleTimer!) and I'm not quite sure how to analyze it. I'm pretty sure I know what aspects I want to analyze, though, hence my being here to share the idea.

First, a productivity heatmap basically shows blocks of time, with each block's colour based on how productive you usually are in that block of time (as measured over many weeks). Keeping it simple for now, we could say that the brighter the colour, the more often [insert productive activity] happens in that time block.

There are several types of timescales to consider for the heatmaps:

• displaying the hours of an average day
  
• displaying the hours of an average week
  

Some people will probably find themselves using one or the other of the above, and some people may find that one or the other is redundant, but I'd bet that many people would like to see both timescales.

(There is one step beyond this that might be utterly fascinating: the ability to check for cycles longer than a week, like months or lunar cycles or seasons or whatever else might be interesting to someone to examine for trends. But, to start, the above two ideas would be plenty.)

There are also several scopes of activities to consider for the heatmaps:

• per activity (so we can examine heatmaps of individual activities)
  
• per category (once they're implemented)
  
• comparing multiple activities and/or categories (in fact, comparing an activity to a category should be possible as well, as I will describe below in an example)
  
• an amalgamation of information, calculated from all (or just a subset) of all recorded activity times, in order to give some kind of overall productivity report (this was discussed in another thread here).

The reason I'm feeling so passionate about this is because I've recently been realizing that it's not actually useful to know how much time I've spent doing a particular activity. That's not why I'm tracking time; I'm tracking so I can find a way to improve, and just knowing the numbers isn't quite inspirational enough. I need to figure out how I tick, and how I can get myself to tick better.

I think this will happen if I can figure out what time of day (and/or the time on a particular day of the week) seems to be the most effective time to spend on my most important (or my most creative, or my most physically or mentally demanding, etc.) tasks.

I think that comparing heatmaps between two different activities (or categories) could help me figure out correlations, which might imply causation (or it might not, since correlation does not imply causation!), or a set of correlating results could be due to some other cause I haven't realized.

An example that I would love to explore is: how does my productive time (which may be several different activities) depend on how much sleep I've had? I mean, yeah, of course it does, but by how much? What is the ideal amount of sleep? My goal is 6-10 hours, but if I could narrow it down more, I'd probably be healthier and more productive. What about naps: am I as productive after two 4-hour naps as I am after 8 hours of sleep? What about exercise? What about time to recharge, away from the computer but awake and doing things (which may be several different activities) that benefit me?

What is important to note here is that I just think it would be useful to visualize the relationships. More views = more understanding of what the data is saying. It can also mean that you don't need to make up some kind of grading scheme, and instead just let people see the data in another way. Of course, there is certainly a way to analyze success based on how the goals are set up, and that could be shown in a heatmap as well, but I still think that the raw data could be translated into visual data that we could usefully interpret for ourselves in an intuitive way.

I've contemplated figuring out how to do the above myself from the exported data by using a spreadsheet, but I couldn't figure out how to just count whether a particular time slot was assigned a particular activity, let alone how to do the rest of what I want. It seems I almost need the data to be in the form that it is originally recorded by me in the bubble sheet: yes/no, is a bubble there or not?

BONUS: It wouldn't even matter if there was multitasking or skipped time slots, because all that's important is counting how many times a particular bubble shows up bubbled!

Have I rambled enough? Do I need a mockup? :)