Applied Nate Silver – another disappointment

This is beginning to be the whole story of this saga, one miss after another. How long until I don’t even see negatives at all? Plus it tells me all the predictive tools are failing to predict the actual result and that means I have no “real-time” data to look at to keep me on my projected path. But perhaps I’ve got a few clues how to adjust and make good predictions again:

weekly15-money

I just have to face it, the more recent trendline doesn’t have as steep a slope as the earlier one. IOW, my weekly losses are slowing down and this week, at only 1.6lbs is the worst yet. This is in line, however, with the idea that the BodyMedia system must be discounted by about 2/3rds (the same as last week). Its “raw” result would have predicted 2.6lbs (calories-burned – calories-consumed) and so I have to discount it only 62% to explain the actual value. Even worse my predicting technique (sedentary calories burned + exercise calories – consumed) predicted 3.5lbs/loss, well behind 2X overstatement.

So let’s add:

  • 10% fudge factor upwards in calories
  • use 70% of predicted loss (after the fudge)
  • adjust all the machines (see below)
  • try to emphasis the hard workouts while going for miles on bike

So something is seriously wrong with predicting now by using various estimates of calories burned vs consumed. In fact, if I ignore exercise entirely (IOW, view last week as sitting on my ass all week) I end up with 1.1lb/loss, which therefore rates my exercise as only 257 calories/day. Despite some discounting I am prepared to do I just can’t believe that.

So what this really means is that I’m fairly drastically under-estimating my calories consumed and that’s somewhat realistic. Last week I had two restaurant meals (pure guess as to calories consumed) and one large at-home meal. Meanwhile the other four days I’d consider fairly accurate, say no more than 10% understated. That means I’m off by about 1.5X on those “big” meals and therefore that’s my real problem on that side of the ledger.

But I also have the big discrepancy between BodyMedia’s measurement of calories burned and my machines’ values as they compute it. I have determined BodyMedia just doesn’t count stationary biking at all and after more testing I believe that system is just glorified pedometer. The additional sensors probably just scale what the pedometer is showing (the problem with pedometers not distinguishing between hard running and easy walking). So with no steps recorded, whatever readings the other three sensors are showing are ignored. This significantly reduces the value of this system which means I have to find another way to adjust.

Now despite liking burning up miles on stationary bike for other reasons it isn’t doing me much good on weight loss. I need to get my own calibration and maybe I can do that. All three machines (treadmill, bike, and elliptical) have heart-rate monitors but they’re all a little erratic and fairly inconsistent with each others. It turns out I may have access to a body mounted HRC I could use on all three and thus connect METs on treadmill, watts on bike and just calories on elliptical.

But then I realized I’ve got another issue with the readout from the machines. All these machines use weight as part of their formula to compute calories and the weights are now out of date. In order to maintain consistency in my stats I’ve adjusted the direct readings but now I realize I should just reset the machines (will do on 1Feb). I’ve got the elliptical, for instance, set at 265lbs and should crank that down to 210. While treadmill is 181, I adjust that by 1.35x to get results for 265lbs. So let’s just reset all machines to 210 on 1Feb and therefore recognize all my records (over 2.5+ years) are now useless, but right now the more critical thing is accurate readings.

Then, if I can get good heart-rate readings, to calibrate across machines, I may be able to come up with factors (for instance, I know I should discount bike a lot) to get comparable values across machines and thus overall get a more accurate reading. BUT, when I’m done with all this it’s still not going to work out to the miss I’ve seen in the last two weeks, so therefore I have to just admit consumption is the bigger issue.

Subjectively I know eating has crept up. The disappointing thing is that the real (as opposed to my too-low estimates) means I’m already around “maintenance” (zero loss) consumption now. And that’s really depressing because I am hungry all the time now (by itself, a good thing, as I interpret that as drop in blood readings as my perceived sense of being hungry is tied to that).

Now all the formula tell me that I can’t be looking at a straight line anyway. Calories burned in exercise drops with lower weight and calorie requirements (just being alive) also drop with weight, so there is a double whammy. Sounds non-linear to me, so let’s take a look.

weekly15-polyProject

The difference is visually subtle but the formula shows the second order effect that is reducing my week loss and this is consistent with the model of calorie burn and requirements. This scale is a little hard to look at for prediction so let’s blow it up a bit.

weekly15-predict

So this is predicting 2.4lbs loss next week and finally reaching 204.9 in four weeks (average 2.2lbs/week). Given the last two weeks those targets are ambitious so I’m going to crank it down (1.9 next week, 1.6 for four weeks), or

211.7 (next Sunday)

(hopefully I can avoid “big” meals this week) and

205 (for week 20)

which does mean 50lbs total in 20 weeks and probably breaking 200 just barely by April (that’s a long time to hold the resolve).

So check back and see how I did.

About these ads

About dmill96

old fat (but now getting trim and fit) guy, who used to create software in Silicon Valley (almost before it was called that), who used to go backpacking and bicycling and cross-country skiing and now geodashes, drives AWD in Wyoming, takes pictures, and writes long blog posts and does xizquvjyk.
This entry was posted in musing, prediction, whine and tagged , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s