For the nerds in you! TDEE estimation, with graphs :)

Hi there again!

I'm a hopeless data lover - I believe the best way to deal with a problem is to study and understand it, because then you can control it, and I've been approaching weight loss the same way - by weighing myself daily and keeping an accurate journal (to the best of my abilities).

I've already posted some graphs on daily weight fluctuations, plateaus etc., and if you're still upset with your scales or get discouraged by what they show, you might want to check that post: http://www.myfitnesspal.com/topics/show/1098806-newbie-loss-plateaus-and-weight-loss-math-with-graphs

Lately I've been trying to avoid work, so I turned to my weight loss data again, and produced a lot of very nerdy plots related to TDEE estimation from actual data. Hopefully it may give some of you ideas on how to kill your time, or help you figure out what's wrong if things don't work as well as planned.

So here we go. I have this excel file that looks like this:

Stuff in boldface is entered manually (daily weigh-ins, calories in, calories out, workout duration), while the rest is automatically calculated as follows:

1. Net calories = calories in - calories out

2. BMR (I used Mifflin St Jeor formula, you're free to put whichever you like) = 10*(weight in kg) + 6.25*(height in cm)-5*age+s, where s=-161 for women and +5 for men. The weight there is actually the "Expected Real weight" column defined lower. This is to compensate for the fact that BMR drops in time, as we lose weight.

3. Net Corrected = Net calories + [BMR column]/1440*[Workout duration column], where 1440 is the minutes in a day.
(This is basically to subtract the equivalent of BMR for the time of your workouts, as HMRs / MFP estimates do not do so, and you're calculating your BMR calories during workout twice)

4. TDEE sedentary (without workout) = [BMR column] * TDEE coefficient cell (which is what we're trying to calculate, I started with the default of 120%, corresponding to "sedentary")

5. Deficit = TDEE sedentary - Net Corrected

6. Expected weight loss = Deficit/7700, where 7700 is the amount of calories needed to lose 1 kg of fat

7. Expected "real" weight = (expected "real" weight of the previous day) - (expected weight loss of the previous day). The first value in this column is set manually, and as a default I took the initial weight the normal linear trend (right click -> add trendline) indicates (another small thing we can try to estimate!)

8. Expected "scale" weight = Expected "real" weight + c, where c is a constant estimating average amount of weight in your stomach. This one is really just for crazy nerds like me:P

With this setup, we can already try to see if our expected loss matches our real loss:

By varying the TDEE coefficient we can obtain several expected weight estimates. Once you hit a line that sort of fits - that's about it!

But if you're nitpicky like me, or you haven't been very consistent with your net calories (you can "identify" that by looking at how "straight" the estimated "real" weight line is - mine for example is straight enough), then you can go further and get the ACTUAL relationship between expected weight and real weight, by generating an XY scatterplot:

Add a trendline, go to "Format trendline" and tick "display equation" and "R value".
Now what you want is to get as close as possible to y(actual weight)=x(expected weight). If y > 1x+…, decrease the TDEE coeff, if y<1x+…, increase it. Play around until it hits y=1x+… If the coefficients get close but the "real" weight starts being consistently too high (or too low), readjust the initial "real" weight, to line it up as nicely as possible.
Finally, if you're really bored, you can calculate the expected "Scale" weight (what you're likely to see on the scales, not the weight of your empty body) by playing with the coefficient in column 8 until you hit y=1x+0, like this:


And here's how they look on the initial graph:


There. Now I need to get back to do some real work!