Weighted Averages for Weight (+ bonus question)

[timc]

I have a running log of over 700 daily entries. Two data points are saved, scale weight (minimum) and kcalories (estimated). These are in a spreadsheet file.

 

"Does anybody use an average or weighted average to compute their 'weight'"???

 

If so, what kind, how many data points do you think are required?

 

A follow up question would be whether there are standard or biologically justifiable approaches to this question. My motivation is that the scale weight is far too variable and what I'm really interested in is body mass minus the variables of water, waste, food. My fitness level should remain similar over time but it is one glaring hole in my approach.

 

I used to use an unweighted 30 day moving average but now I use a weighted 28 day moving average such that the latest data point is worth ten times more than the oldest (linear in between those two for days 2-27).

 

Having NOT done this in the past, I will say it a several times more useful because it takes the ever fluctuating thing known as 'weight' and turns it into a useful, achievable metric. The scale number can be several pounds higher while the result - the average - is on target. Only multiple days at the feed trough would affect things negatively. There is also the achievable goal of being under a certain calorie level, but I'm more focused on the result (and specifically not altering that too quickly).

 

More so, this is the only way I know to narrow down on the desired calorie level. A standard approach of ad lib minus ten to thirty percent will not work for me as I don't know my ad lib results. If I measure or weigh, then my weight and consumption goes down. If I don't give a F, then nobody is recording the results (maybe the NSA could total up my food receipts from years ago).

 

Additional to not knowing my ad lib results, my years of observation suggest I'll want to end up in the 18-22 BMI range (high side of that range more likely or a VERY slow crawl down that range) with a calorie rate under 2500 (presently, I am under 2600 kcal/day on average). Unless disease or advancing age refocused my efforts, of course ...

 

E.g., presently, I'm exceeding my goal of ~1 lb/month and am trying to slow down the rate (not too hard of course) and I may soon try to target in on 0.5 lbs/month or less. The method also creates smoother charts showing the relationship between (average) intake and (average) weight. Otherwise my calorie data points would jump 1000-5000 kcal and the weight data points as much as 8 lbs (typical day-to-day variation is only a few pounds).

 

 

 

A follow up question, if I were to add a single metric for fitness, what ought that be? Tape measurements (waist? hip?) or maybe one of those fat-estimating scales??? If it actually involves exercise, that could be interesting (like a weekly Presidential Fitness Test style thing) and it would be the only formal exercise I do presently that isn't running around at work or for soccer.

[BrianMDelaney]

Hey Tim,

 

I've used 1) a weekly and 2) a monthly avg. for years, but this isn't so important to me now, since I'm not trying to change my overall weight target (i.e., "progress" means something different to me) – rather, I'm using weight as a proxy for daily, "effective" caloric intake ("effective" because of gut issues: counting calories isn't enough).

 

As for a single measure of fitness -- tricky, since there are so many ways to be fit. But the Presidential Fitness test idea sounds great. One alternative I used to use is resting pulse. For a given, fairly low (but not extreme), stable level of energy-intake, morning resting pulse of 43 meant extremely aerobically conditioned, 49 meant out of shape (for this moderate-ish CRer). It was a surprisingly useful metric (taken as an avg.). 45 really meant something different from 43, for ex.

 

Brian

[nicholson]

Hi Tim: 

 

As regards measuring weight what I have done for years is not take averages of any kind but instead to weigh myself on a daily basis, at the same time every mornng after getting up and emptying my bladder, plot the data on graph paper and simply draw a line by eye through the middle of all these observations.  That line represents, imo, my real weight.  The range between the highest and lowest of these observations is commonly six pounds - three pounds both above and below what I take to be my real weight.  The main contributors to the width of the range are, as you note, intestinal contents and also SALT intake.  High salt intake causes high salt absorption, which creates thirst as the body attempts to maintain electrolyte homeostasis by dilution of salt content with water.  Eventually the salt is excreted in urine and water is lost and weight returns to the 'normal' mid-line.

 

As regards 'fitness' I suggest measuring out exactly one mile on a flat surface suitable for walking, and then walking it at the fastest pace you can manage.  No running permitted - at least one foot must be on the ground at all times.  There are (or used to be) some data in the files over at CRsupportgroup with people's walking times (including mine!) over one mile.  IMO, when doing this it is very helpful to record your resting pulse (perhaps in bed before getting up in the morning), your pulse rate at the end of the mile walk, and pulse rate two minutes after finishing the mile (speed of recovery).  Each of these are pretty interesting measures of what I regard fitness to be.

   

Rodney.

[nicholson]

Regarding my mention of salt intake above:  Oriental recipes contain huge amounts of salt (perhaps mostly from soy sauce?) and any time I eat at an oriental restaurant my weight jumps a few to several pounds as a result of water weight, and then subsides over the next few days as the salt is gradually excreted.  So, to clarify, I do not eat large amounts of salt on a regular basis.  JFTR the largest 24 hour increase in body weight I have ever experienced as a result of ingestion of salt was 11 pounds!  NOT joking.

 

For anyone curious about the effects of ingesting larger amounts of salt, a simple way would be to take a look at the leading causes of death in mainland China.  Stroke accounts for about one quarter of all deaths in China, and stomach cancer is a much greater percentage of deaths than over here.  In contrast, and a point worth digesting, imo, coronary heart disease is almost negligible in China.  So there appears to be no justification for concern about salt causing heart disease.

 

Rodney.

[timc]

Hi Brian,

 

Thanks for the resting pulse suggestion. I ordered this pulsemeter cheap - I lack the attention span to do it with a timer - with Amazon prime shipping:

 

http://amzn.com/B00SCPBKFO

 

I'll also try to dig out a tape measurerer (fabric for waist/hip ratio or something).

 

 

 

As regards measuring weight what I have done for years is not take averages of any kind but instead to weigh myself on a daily basis, at the same time every mornng after getting up and emptying my bladder, plot the data on graph paper and simply draw a line by eye through the middle of all these observations.  That line represents, imo, my real weight.

Hi Rodney,

 

I think that is somewhat of an average as well just done by sight as you suggest. I've noticed gains too after lots of exercise/dehydration. The body compensates by storing more water even if there is no intent to repeat the same routine.

 

My schedule is far too random to use the same time of day. Too often, I'm rushing off to work shortly after waking up but usually at a diferent times thoughout the week (I wake up at least 3 or 4 different times based on my schedule).

[BrianMDelaney]

Tim, wow, the price is certainly right on that model ($17.49). And it's an oximeter also. The only downside is it would make typing difficult. But if I'm going to be reading or doing something else, it might be ideal for me. I'm going to read some reviews of other brands and see how they compare. A little more money for extra features (infrared transfer of data, for ex.) or quality might be worth it.

 

Brian

[Dean Pomerleau]

Brian,

 

If all you want is a good way to measure resting heart rate, there are good ones for both Android and iPhone. I'm particularly fond of the free Instant Heart Rate app for Android (not sure if there is an iPhone version). It uses the phone's flash and camera to monitor your heart rate and is quite accurate I've found.

 

But have you considered at fitness tracker rather than a dedicated HR monitor? I really like the Fitbit Charge HR, which is $137 from Amazon.

 

It provides lots of data in addition to heart rate, including 24 hour tracking of heart rate and sleep time (not sleep stage, but amount of time spent asleep, restless and awake at night). It has a really nice web-based and app-based dashboard for visualizing the data. Here are a few examples from the web-based dashboard.

 

First, here is my 5-min by 5-min heart rate throughout the day yesterday:

 

 

The peaks above 150 BPM aren't accurate. The reason they are so high is that during those times I was on my stationary bike(s) wearing my Fitbit around my calf to count steps. Its only accurate for HR when worn on the wrist, and tends to grossly overestimate HR when worn on my leg. My HR during those bouts of cycling ranges between 90 and 110, and the Fitbit is quite accurate about tracking those readings if I wear it on my wrist rather than my calf, but then it doesn't count steps (= pedal revolutions).

 

The other interesting thing to note above is that you can hover your mouse over any point to see the exact value, e.g. the 41 BPM shown above for 2:40-2:50am. That is quite accurate.

 

The Fitbit also uses a algorithm to estimate your resting heart rate, which they consider to be your heart rate when you are inactive but not sleeping. You can see resting HR in the form of a long-term graph, which I find very informative. Here is an example:

 

 

You can see today my resting heart rate is 50 BMP, and has varied from 46 to 53 BPM over the last month. It definitely correlates with calorie intake.

 

Sleep is another thing the Fitbit tracks, which I find very useful. Here is a graphic of my sleep pattern over the last 7 days:

 

 

As you can see it tracks sleep time, restless time/count and awake time/count during the night. It automatically detects when you go to bed and get up using its motion sensors. Last night I was asleep for 6hr:26min, awake for 11min, and restless for 27min. As you can see I'm a morning person (8am bedtime, 3am rise time). It quite accurately counts the number of times I get up in the night, usually to go to the bathroom.

 

You can also see sleep over the last 30 days:

 

 

By hovering over a bar you can see its details. As you can see, Aug 19th was a day when I woke up early, and only got 5.28 hours of sleep, but most days I get around 6.5 hours.

 

The other thing I track is step count, which includes running, walking and cycling (one step per revolution of the pedals when I wear the Fitbit around my calf). Here is a graph of daily step total for the last week (Saturday is partial):

 

 

I spend a lot of time on my stationary bike(s), so the steps are pretty high. I've verified the counts are quite accurate for me (within a few percent).

 

Anyway, I've found the data from my Fitbit to be quite useful, particularly for tracking resting HR, sleep and steps.

 

--Dean

[James Cain]

have you considered at fitness tracker rather than a dedicated HR monitor? I really like the Fitbit Charge HR, which is $137 from Amazon.

 

....

 

I spend a lot of time on my stationary bike(s), so the steps are pretty high. I've verified the counts are quite accurate for me (within a few percent).

 

Anyway, I've found the data from my Fitbit to be quite useful, particularly for tracking resting HR, sleep and steps.

 

--Dean

I'll mirror Dean's experiences and enthusiastic recommendation for the Fitbit Charge HR. Both my wife and I got one (she actually got the next model up with the built-in GPS) a few months back and have loved them. Mostly very accurate for all of the above measures, and their algorithms for any not-directly-measured calculations seem spot on. For example, it knows when I'm 'active,' when I fall asleep, miles walked based on steps (I think based on step count, frequency, and intensity), etc.

[BrianMDelaney]

Wow, these are some of the most glowing reviews of Fitbit I've seen.

 

I had looked into fitness trackers, but decided I would wait 1-2 years for the tech. to mature a bit. But maybe it's worth it to purchase it now, even knowing that in a year or two something much better will come along.

 

But one has to pay more to export the data?

 

http://www.fitbit.com/premium/export

 

That would be a bit irritating. I'm assuming I'll be buying a tracker from a different company at some point, and will then want to have my data from the Fitbit -- all of it -- in order to merge it with the data from the new tracker.

 

Actually, having to choose between accurate "step" (or pedal) count and accurate pulse measuring gives me pause. When will an affordable dual-sensor (or triple-sensor ...) fitness tracker be available? I suppose, for now, I could use my heart-rate monitor while cycling and put the fitbit on my ankle, as long as I had access to my Fitbit data and could edit in the heart rate data from the heart-rate monitor with a Word or Excel macro or something. Now I'm remembering why I decided against getting a fitness tracker in this early phase of the "S-curve" of tech. maturation.

 

- Brian

[Dean Pomerleau]

I just played around with the fitbit export feature. It appears that for free you can export a month of daily total data at a time - not just the last month but any month, in csv or xls format. If you want more than a month at a time, you need to upgrade to premium for $50 per year. I presume you can upgrade at the time you want to do a big export, so you could think of it as a one time fee to export your data in one fell swoop, rather than the tedious month at a time way.

 

Not ideal, but at least there are options.

 

Dean

[Dean Pomerleau]

I've found my Fitbit Charge HR to accurately track HR and steps (as well as distance) when I'm walking or jogging when mounted on my wrist. It's only for cycling or pushing a shopping cart, when your arms aren't moving, that the fitbit can't count both very accurately.

 

I have a chest strap HR monitor to spot check my HR while on my road bike, and the new Exepeutic stationary bike as a HR monitor built into the handles.

 

I find at my usual pace my HR doesn't vary much while cycling, so it not necessary to track it continuously.

 

Dean

[BrianMDelaney]

Dean, thanks for doing more experimentation and letting us know. Looks like Fitbit is excellent for tracking most of what I want to track.

 

Not ideal, but at least there are options.

 

The sticking point for me is having to merge, or fudge, data on a regular basis. My needs are maybe a bit different from others'. One of the things I really want to be able to do is have as much data as possible together in a format that allows a program to search for patterns ("big data" mining, as people now, often sloppily, are saying). Doing a monthly download & merge into my one big file wouldn't be a problem. It's only once a month. But having to merge heartrate data with step data on a daily basis would be irritating.

 

Amiigo might solve this. I'm going to read more reviews and report back later.

 

- Brian

 

[timc]

Tim, wow, the price is certainly right on that model ($17.49). And it's an oximeter also. The only downside is it would make typing difficult. But if I'm going to be reading or doing something else, it might be ideal for me. I'm going to read some reviews of other brands and see how they compare. A little more money for extra features (infrared transfer of data, for ex.) or quality might be worth it.

 

Brian

Regarding this pulse gizmo:

http://amzn.com/B00SCPBKFO

 

I have used the meter to track my low BPM for two weeks (average 54 BPM, oximeter 96.8 - which I haven't researched what that means). No, I wouldn't try to type with it on for sure.

 

There are no trends to report but based on the numbers above, I would expect it to decrease with weightloss.

 

 

The idea of constant tracking (like a Fitbit) is possibly overwhelming to my wee brain. I'd want to see how well it estimates calorie expenditure (total) along with the other data to give meaningful totals/averages for each day. Also, I'm not sure how the ergonomics of the tech will evolve. For some reason and ankle monitor would seem preferable as I'm both watchless and would need soccer refs to clear its use (there are 'no jewelry' rules - a picky ref would mean no monitoring exactly when I need it the most). Results for the Fitbit on ankles as Dean mentioned are variable and others report weirdness (like doubling calorie expenditure for the same workout).

 

With ankle monitoring, nobody would see it and if they did I'd only have to convince them that I'm free to leave the county without issue.

 

How well does Fitbit calorie tracking match up with the food shoveled in? Obviously if one's weight is constant and the monitoring is constant and the calorie counting on target, these should line up well.

 

Most important of all, I'd wait for one without a privacy policy page:

http://www.fitbit.com/privacy

(it links to other privacy pages if you manage to read far enough into it - this makes me wonder if those privacy pages link to even more privacy pages ... and yes they do: https://www.adroll.com/about/privacy

 

[by way of example, the pulse monitor I bought doesn't need a privacy page.]

[Dean Pomerleau]

Tim,

 

I don't trust the accuracy of calorie expenditure for any of the available fitness trackers, but by incorporating pulse I think the Fitbit Charge HR is more accurate than most, except for the problem with no arm movements during cycling.

 

I don't trust estimates of calorie INPUT either, that's why I eat nearly the same thing everyday and track my weight instead of calories expended / consumed.

 

Regarding privacy, yes you give up a lot on that front in order to get data in an easily usable format with today's fitness trackers. If you're concerned about that sort of thing, best to stay away for now.

 

What about CRONometer? Are you OK with its privacy policy?

 

http://cronometer.com/privacy.jsp

 

Dean

[timc]

What about CRONometer? Are you OK with its privacy policy?

 

http://cronometer.com/privacy.jsp

 

Dean

It's certainly preferable to most (and not as open ended with the ads being seemingly limited to 3rd party banner ads - not sent to email).

 

Regarding cronometer, I mostly use the USDA website or just a general search when looking up calorie estimates (which I would expect even averaged might be a +/- 10% or more). Any data is on a local spreadsheet (or my cell phone until I transfer it). And then copy/paste of identical meals helps. And the kitchen scale is a huge help. Often items don't match their package weight so adjustments can be made.

 

Regarding future devices, if a bluetooth gizmo had temperature, pulse and oximeter - maybe it could get a closer expenditure estimate (for a given height, weight, age, sex).