I’ve done a number of park runs now, and they are kind enough to give you various bits of data surrounding how you’ve done, and I thought it would be cool to see how this data looks visually.
I have used age grade rather than time because a high age grade is faster and it makes more sense to say higher is better (age grade is a measure park run uses to see how good your time is compared to other people of your age and gender, and is measured as a percentage. 25 minutes for a 21 year old man will have a lower age grade percentage than 25 minutes for a 55 year old woman).
Much as I tried to change it, I could not work out how to get Google Sheets to use a sensible date format, so 7/1 here means 1st of July. Anyway, here is how my park run results have changed over time.
The blue dots are the actual times, while the red dots are the median times from the previous 8 results on and before the date in question. The median is simply the middle value in a set of data (for example, the median of the set of numbers 1-9 is 5). I am plotting the median as well because it means that very fast or very slow runs will still increase or decrease the value, but not in a way that causes the trend to be particularly spiky.
My first run was on New Year’s Eve 2016. 2017 seemed to show a general downward trend, which is not so easily explained. In April of that year, I was challenged to run 5k every day (except Sundays), during which time I got my (then) worst time at Portobello, and the following week, got my (then) best time at Cramond, an agonising 3 seconds away from my goal of getting under 25 minutes. I don’t know if another near-miss the following week disheartened me, because after that my times seemed to go down the rest of the year (with a mini-revival in the summer), hitting a low on 28th October (an interesting park run, which was Halloween themed, and being chased by death did not seem to make me go any quicker).
I’ve been getting steadily quicker since then, I think because of deliberately making an effort to do longer distances during the week. In February, I signed up to do the Edinburgh Half Marathon, and have been doing at least one 10k+ run per week; since then I have not gone over 27 minutes.
The R^2 of 0.348 means that it looks like the trendline is an ok estimate of how I’ve been doing, but that it could be a lot better (i.e. it might be a bit much to say definitively that my performance got better, worse, then better again, but it’s a decent-ish estimate for what we have). The R^2 for the median is 0.995, meaning it’s an excellent estimate,
I thought it would be worth breaking it down by location, as below.
Annoyingly, the legend has been cut off, but blue represents my Cramond times, red Portobello, and yellow others (twice in Belfast, once in Livingston).
I haven’t included a trendline here because it had an R^2 of zero (which basically means there is no trend in individual courses, which is quite revealing in itself). I frequently describe Portobello as being harder than Cramond, but this doesn’t seem to bear out too much in the data (though there’s a caveat that maybe if I was putting as much effort in at Cramond as I was at Portobello in 2018, then maybe those times would be even better). Portobello is narrower, has a few minor hills, is more cramped at the start, and has about 10 times as many tight turns as Cramond does (indeed, Cramond only has two turns worth talking about).
The more I do, the better the estimates will get, so there’s an incentive.