Message Boards » Website Suggestions/Feedback

TOPIC: Moving average weight

 
Ic_disabled_photos
Topic has been inactive for 30 days or more and images have been disabled.
Display All Images
August 9, 2011 4:38 AM
Hi all

It's often recognised that measuring weight too often and focusing on it is detrimental. People get overwhelmed by small day-to-day fluctuation and don't see the big picture.

While weighing weekly is a good solution, many people struggle to stick to that. (Reminder notifications would be nice for that, BTW).

What I'd love to see would be MFP support a moving average weight over the last 5 or 7 days, whatever. It'd show the individual entry data points, but the line would be a moving average curve. That'd damp out daily fluctuation and let people who feel compelled to weigh daily to still get a better measure of what it actually means ... and *SEE* how noisy/error prone the individual daily measurements really are on the graph.

Eating distorder and food health psychologists everywhere would love you.

It's not actually a problem for me - I stick pretty well to weekly weighing - but it'd be a big help for many.

For anyone wondering about moving averages: http://en.wikipedia.org/wiki/Moving_average

... and here's a sample graph I whipped up that shows the difference between a 4-day moving average and individual samples. There are probably better algorithms than a moving average, as a moving average tends to "lag" behind, but it gives the general idea. Perhaps using a moving average over the past n datapoints *and* the n next extrapolated datapoints would do better. Anyway, what I want is something like this:

Image not displayed
Edited by amoonymous On August 9, 2011 5:25 AM
August 9, 2011 4:45 AM
I totally agree with you, it would be a help to me also
  2793411
August 9, 2011 5:53 AM
I do this on an excel spreadsheet and started doing this about a month ago ... it has been eye opening because even when I dont "lose weight" ona given day persay, the average has consistently gone down and that is fun to see.
  2545488
August 9, 2011 6:02 AM
emmaleigh47: Yep, and in fact I generated the graph above with OpenOffice Calc (like Excel, but $0). Good plan. It'd be nice if we could export the logged weight data as CSV from MFP to make this easier and avoid double-entry of data, though.

Because I only weigh weekly I generated the above graph by hand-re-entering my weekly weights, then generating several intermediate data points between each measurement by introducing random noise. It replicates the kind of results you'd see from real every-day weighing, though, and matches what I've observed when I've done more frequent weighing in the past.
March 19, 2012 7:41 AM
An exponential moving average (EMA) really is the ideal way to track your weight. I started using the Hacker's Diet site to track my weight about a year ago and it provides a much more accurate measure of your actual progress toward your goal. The nature of a moving average means that your calculated weight always lags your true weight just a little, but it smooths out all of the day-to-day variation and gives you a real view of what your weight is doing, while still allowing you to weigh in as frequently as you like.

I like to weigh in daily, at least in part because I enjoy seeing the daily downward trend of my weight. As emmaleigh47 said, with a one or two-week EMA, each daily weigh in tends to pull the average down a bit, even if the measured weight happens to be a little higher than the previous day's weigh-in. That daily reinforcement of the fact that your diet/exercise program is actually working is really valuable, IMO.

This would be a really nice addition to MFP.

What I've taken to doing is to enter my daily weights on the Hacker's Diet web site and enter the weight it computes for me into MFP, so my MFP weight graph is smooth, and I don't have big fluctuations that cause spurious "You lost X pounds since your last weigh-in!" posts on my stream.

Reply

Message Boards » Website Suggestions/Feedback

Posts by members, moderators and admins should not be considered medical advice and no guarantee is made against accuracy.