Thanks! Yah I thought it would a be fun exercise to create this.
Thanks! Yah I thought it would a be fun exercise to create this.
It's definitely fun.
FWIW, it gave a pretty inaccurate result for mine.
On Friday, I ran a track workout of 3200m, 1600m, splits were 12:46 and 6:09, with 6:49 rest. Based on that workout, it predicts an 18:32 5K for me.
My road PR is 19:10, and about 5 days before I ran a 5K in just over 20 minutes. It was a slow course and I didn't pace my race well, but I still find it very hard to believe I've got an 18:32 5K in me right now, on either the roads or the track.
I'll add a few more workouts in and see if it modifies the prediction.
Hi Darkwave,
Thanks for the feedback. Your prediction/misprediction align with the current state of the model - it gives a bit too much credit for longer repeats. It's something I'd like to address in the next version. I will note that based on my data track 5ks workout to 30s or so faster than road 5ks. The predictions are still a bit fast for you but the different surface and environment could lead to faster times.
Last I saw Canova seemed to disagree with this (track being faster than road) but from what I've seen in the data especially from the general population, all else equal times are faster on the track. Whether is the surface, the conditions, the pacing, or the general atmosphere I can't say.
[quote]Metric vs. Imperial wrote:
2. The intervals need to be entered in Metric and the mileage needs to be entered in Imperial. Is that intentional? quote]
That shouldn't be a problem. Every 5k I have entered the last 35 years has had mile splits. I have never run a 5k that gave kilometer splits.
Seems reasonable.
Very good job, I believe.
I put in a couple of recent, but very different sessions - 12 x 400m with 1 min recovery, and 6 x 1600m with 4 1/2 min recovery and the predicted 5k times were within 10 seconds of each other, and pretty near what I believe I could run at the moment (I'm 58, and have been racing for 40 years+ so would have a pretty good "guestimate").
With machine learning, it's just going to improve as you get more data.
Runners are probably going to get a better result by averaging the prediction two different sessions, as depending on what type of runner you are, you may well excel on one kind, and struggle on another.
I think track is generally faster than road, and it's very rare that you get a perfectly flat road course, so variation from track sessions to road races is likely to be great. It's also somewhat harder to get the pacing right on the road.
Overall, a very useful guide, and at minimum you can at least use it as a trend indicator even if you didn't take the projected times as absolutes.
It seems really off to me. It says i'm in 14:48 shape, but I'm a high school junior whose pr is 16:53.
You probably aren't using it right...
Says I'm in 17:08 shape and have been 16:19 in the last couple of weeks.
Wow, awesome site!
You mention on the web page that publicly available logs have been used. Which sites are the sources of your data? Strava? Garmin connect? did you take any filtering (amatuers, elites, high shool runners, masters, etc... or did you use all the data you can get?) Just curious on what the statistics is based up on and how reliable it is..
but really nice work!:-)
Saw the site. Who's your target audience? I mean, there are a hell of a lot of non-lets-runners who have never done an interval workout, so the calculator would be useless to them. On top of that, there are many lets-runners who haven't done an interval workout within several months. So it seems strange to confine yourself to just interval input data.
That said, I did use a recent interval workout for myself. The result was a prediction a little more than 2 minutes 20 seconds off my 3 recent race-times, and a little more than 2 mintues 20 seconds off my recent solo-ed time-trial on a track. So the thing isn't very accurate for me at all.
It also bothers me that there's so much in lower-case on the prediction page...and the font is not good - the letters are spaced too far apart within words, like with Microsoft Word in very recent versions.
Thank you for putting this up! I love the idea of the calculator but I'm not too sure this is entirely accurate.
Workout 8/21/15
4x1600m w/ 90 sec rest
5:18, 5:09, 4:56, 4:49
13:42 5k
Workout 7/1/15
Tempo Run 6 miles
5:52, 5:46, 5:26, 5:19, 5:23. 5:08
13:53 5k
You know I would love to run sub 14 but I don't think that's happening anytime soon.. I think it gives too much credit but anyways, thanks for posting this. I'll plug in other workouts in the future!
bump. This calculator seems to have stalled in its development.
Tooth wrote:
It seems really off to me. It says i'm in 14:48 shape, but I'm a high school junior whose pr is 16:53.
Yeah I just put in my data and a standard session and it said I was in 13:12 shape. Ha! Venti comeback I was wrong somebody's calculator is worse than yours.
A steller 5k runner wrote:
No offense but Im not going to create an account to find out a 5k time.
Agreed. Would probably give it a shot (and tell others about it) but as soon as I scrolled down and saw I had to "sign up" to try it, I closed the window. Is there a reason you need people to sign up? If you are doing it to get stats on how many people are using it, there are countless tools you can imbed to find out, and most hosts have analytics in their dashboard with tons of stats to look at.
It won't let me enter a date. Otherwise awesome idea and cool site!
5 x 800 2:35 2:00 recovery Prediction is 16:46!!
I'd be lucky to run a 17:46
Evidence of the 20 miles per week and mostly high end sort interval speedwork. The 800's done every couple of weeks is for strength. I can barely run a 5 minute mile and then it drops off very quickly. 2 miles in 11:00.
My long run is only 5.5-6 miles.
If you have a 400/800 runner at 20 miles per week this calculator won't work for them.
I'd say adjust down for the lower weekly mileage runners.
Just figured it out and got my result. Very accurate for me. Awesome!
I think it gives you a little too much credit. I entered 8x400 at 69 with a 60 second rest and it put me at 15:05. That would be great but I'm about a minute slower and I don't feel like I was over doing that workout.
what format is required for the date?
keep getting an error message