Sep 4, 2008 8:47pm

cachapa cachapa
3 posts

Measuring the weight of a person is very error prone. The body weight fluctuates a lot during the day for various reasons, and most home scales aren’t that precise anyway.
For this reason, a daily weight measurement will usually be full of ups and downs, while the real data hides behind the line: the tendency of the line (i.e. going up or down).

John Walker explains the problem brilliantly here: Signal Noise while giving a solution to the problem.
In fact, he has a charting tool available on his website, which is open source: The Hacker’s Diet Online

My suggestion is that Gyminee could adopt a similar implementation for the weight charts that we see in the Locker Room.

 
Sep 4, 2008 9:33pm

OlympicHarvest OlympicHarvest
28 posts

It’s a good idea I think to provide an average curve indeed – especially for the people who tend to compulsively weight themselves. Personally I just weight myself every 3 weeks, to avoid these ‘teeth of terror’ that are amusingly described in the link you posted :)

 
Sep 4, 2008 10:15pm

StevenG StevenG
197 posts

Channel Champion

I use a very low tech approach as I know there may be significant fluctuation on a day-to-day basis, I weigh myself at approximately the same time each day and I only weigh-in when I loose (or gain if that is your goal). It makes for longer plateaus but the visual serves as a positive reinforcement and ends up showing the same trend line more statistically sophisticated tools provide.

If there is a prolonged weight gain or loss (two three weeks) I can then make a decision to enter a new data point to create a more relevant trend line.


Login to Reply