Examination of Outlying Scale Fluctuation

steveko89
steveko89 Posts: 2,214 Member
As I was logging my daily weight and calories into my spreadsheet, like usual, it occurred to me to check to see if I'd encountered any outliers in the 300+ data points I've collected dating back to July 2017. I'd inserted a Median absolute deviation row in my data set months ago; to this day I still only have recorded one weight that fails that outlier check, a weight of 186.8 lbs on Saturday 8/19/2017. Curiosity lead me to look back at the confluence of circumstances as a monstration of both the magnitude of scale fluctuations, the requisite surrounding circumstances, and the visibility afforded by daily weighing and diligent calorie tracking.

*For more information on median absolute deviation see:
https://en.wikipedia.org/wiki/Median_absolute_deviation

Executive Summary (TL;DR):
I once saw a swing of +5.8 lbs one day, and -8.8 lbs the next as a result of higher-than-usual carb/sodium intake followed by exercise and a lower-than-usual carb/sodium intake. Scale weight can be fickle, daily fluctuations vary greatly.

The long and meandering version:
This week was a bit of oddity in itself as I was traveling for work Tuesday-Thursday, one of only three trips I made all of last year, so I was unable to weigh myself on Wednesday or Thursday morning, and ate out for lunch, and dinner Tuesday & Wednesday, and breakfast & lunch on Thursday; certainly somewhat of a limitation to my calorie counts for those days. However, weight and calories had been fairly consistent leading to this point.

Sun 8/13 - 181.0
Mon 8/14 - 180.4
Tue 8/15 - 180.6
Friday 8/18 - 181.0
Moving average 8/18 per Happy Scale: 180.8

I was happy to see that my deviation in routine, schedule, and moderate indulgence in Southern cuisine (visited Lafayette, LA) did not have an overwhelming impact on the scale Friday morning. At the time I was working out 3x/week lifting on a full body workout plan. With the travel planned I did my Monday workout as usual, pushed my Wednesday workout to Thursday afternoon once I returned home on a morning flight, and shifted Friday's workout to Saturday to allow for adequate recovery; so no exercise was done on Friday.

Having been away and somewhat out of sync, that Friday was a bit of a culinary excursion. I had my usual breakfast protein shake and Oikos greek yogurt in the morning but had not adequately planned for lunch, which resulted in obtaining a protein bar, a bag of Smart food popcorn, and a bag of cheddar jalapeno Cheetos from the break room vending machine as lunch. My wife and I can only be described as breakfast-food-enthusiasts, so dinner that night was eggs, pancakes, and goetta (a Cincinnati favorite, hard to explain, kind of like a soft sausage that you pan fry, no oil or butter needed). The tally by the end of the day was 2962 calories, my average TDEE works out to ~2450, so call maintenance sans exercise at 2200. Obviously, this quantity of calories is over maintenance, but not necessarily egregiously. The bigger deviation came from the carbs, I typically keep to 70-100g carbs, however 8/18 saw 236g logged. This figure is probably somewhat low as I realize I logged pancakes but not any syrup... and what sort of animal isn't having syrup with pancakes. This far removed from that day I can only imagine that I decided to skip out on weighing/logging my portion of syrup since I already knew I was over my calories for the day and that 100-200 calories wasn't going to make a huge dent in my end-of-day number... little did I know I would be waxing poetically about said evening almost a year later. Additionally, I racked up 4800+ mg of sodium over the course of the day where I will rarely sniff the default limit for my size of 2300mg.

Sure enough, the next morning I was greeted with 186.8 lbs on the scale. "Well shoot, that's going to screw up my data", I thought... probably. Again, I don't really remember what was going on that day, aside from I know I worked out as planned because I recorded a workout with UA Record, following my typical-for-the-time, full body lifting scheme of Squats, Deadlift, Bench Press, Overhead Press, Rows, and Pull Ups. Looking at the food choices, I'm guessing my wife was out most of the day as I skipped breakfast, more than likely to workout in the morning, and had a 4-egg cheese omelettes for both lunch and dinner, accompanied by a protein shake and two yogurts. That's usually the sort of thing I'll do if I'm left to my own devices. The result was a modest total of 1704 cal, exceeding my 1g/lb protein goal with 184g, only 41g carbs and <1750mg sodium. Unfortunately, I didn't track water either Friday or Saturday, so I cannot provide any context for hydration effects. As a rule I'm generally bad at drinking enough water on the weekends, usually hit 8-10 cups/day while at work.

Sunday morning comes and I weigh in a a svelte 178 lbs!?! Sunday proceeded to be another day around maintenance at 2452, (516 of which were cupcakes...) and was back to my regularly scheduled programming that Monday with a weight of 181.6lbs, beginning a week that was on the front end of a pretty lengthy period of slow, yet fairly sustained loss, reaching a low of 170 lbs on January 20th, 2018.

Overall, I don't really know how valuable anyone will find any of this information but I thought it was interesting that the surrounding circumstances of that outlying data point were relatively mild in what, and how much I consumed. It wasn't like I went on a fast food-fueled bender, then fasted and ran a triathlon to sweat that many pounds away. It's entirely likely I didn't even go outside that Saturday, if I did it was probably for about an hour to cut the grass, not exactly an arduous task that would have a tremendous impact on my NEAT for the day.

Replies

  • stanmann571
    stanmann571 Posts: 5,728 Member
    That's really a great writeup.
  • AnnPT77
    AnnPT77 Posts: 31,953 Member
    Data geek swoon happening over here. Good stuff!

    (The food reporting meander is a hoot, too: Bonus points. :drinker: ).
  • AnnPT77
    AnnPT77 Posts: 31,953 Member
    Y'know, this really is worthy of a bump. ;)
  • quiksylver296
    quiksylver296 Posts: 28,442 Member
    This is beautifully geeky.
  • karris43
    karris43 Posts: 34 Member
    Thank you for this. You have inspired me to record more data aka get to know my body.
  • Phirrgus
    Phirrgus Posts: 1,894 Member
    I enjoyed reading that, thanks.
  • PAV8888
    PAV8888 Posts: 13,538 Member
    edited February 2020
    So 10 day trailing average (aka, I believe, trendweight) works as advertised? 🤔 AND you win the weigh in data points contest! 😎
  • steveko89
    steveko89 Posts: 2,214 Member
    edited August 2021
    Linked this in another thread again so I thought it was time for an update. I'm up over four full years of data now and still only two true outliers.

    bmo8cv9xa6dg.png

    In the context of observing trend weight with respect to water fluctuations (as per the question in the other thread), the smoothing effect is predictably far more pronounced on the peaks rather than the valleys, and the average is virtually identical (0.01% difference). The max trend weight is 2.83% lower than max daily weight while my min trend is 0.87% higher than the daily min number.