Why?
Scratching my own itch! My long runs are usually based on time rather than distance, but ultimately I don't want to end up miles from home because my chosen route was too long, or end up doing laps of local streets because it was too short.
Strava and Garmin each give estimates when planning routes, but I don't find these particularly accurate, and I don't know how their algorithms work. I wanted to ensure that 1 hour at 1 metre elevation is treated differently to 1 hour at 500 metres elevation.
How does it work?
When you submit your run data, the service uses a multivariate linear regression algorithm to calculate a "best-fit" line across each of the runs, taking into account time, elevation and distance. Once the model is "trained", we can ask it questions, in this case "how long will a run be, given a certain time and elevation.
This will work best when inputting long runs only, that you run at a consistent pace. It will be less accurate if you include intervals/tempo sessions.
Technically this is machine learning, so I'm hoping for vast amounts of VC money to be thrown my way any day now...
What about privacy? are you stealing all my run data?
Honestly I'm far too lazy for all of that. Your run data is submitted to a server, which uses it to train a regression model. Once that is done, and the prediciton is made, the data is discarded. For ease of repeated-use, the site uses your browser's local storage to retain run history. You can clear this using a button on the main page.
Who are you?
This site is made by Lee Morris, a software developer from High Wycombe in the UK.
Can I get in touch?
Sure! I'm happy to receive feature requests and bug reports. Email me at [email protected]