Statistics approach

One of the main values I get from Beeminder is the graphs and data tracking. It’s a fantastic one-stop shop for all my quantified-self monitoring, with all the IFTTT and auto-logging features. But what I’m missing is analysis of this data - the statistics section just isn’t useful to me in most cases.

Currently it tracks total number of datapoints (generally not what I’m interested in), average value (in many cases the y-axis number of my graphs is irrelevant), and mean delta. The mean delta is closest to what I’m looking for, but it’s averaged over the entire life of the goal, which means that if I’ve made changes, or archived it for a while, this number becomes irrelevant too.

I’d really like to see statistics that are broken out over certain time periods. Trendweight ( does this fantastically for weight tracking - it gives deltas for the last day, week, two weeks, and month, so you can see both the first and second derivative of your progress. For Beeminder I think the delta over the past day, week, month, and year would probably cover most use cases, especially if it was windowed by the chosen x-axis settings. I’d love to see, for example, how often I’ve been to the gym over various time periods, or how much sleep I’ve been getting.