Analysis of the past month (including charts and data)

dengarrett

Thanks for the tip on the moving average - curious as to why you picked 7 days?

Good question. Unfortunately the answer is "it feels about right"

It's sufficiently responsive that you don't miss trends when they start, but sufficiently smoothed to hide fluctuations?

It also ties in with the advice to weigh once a week (which is a very crude 7 day average.....)

As playing with numbers is your comfort zone I'll recommend the hackers diet to you - http://www.fourmilab.ch/hackdiet/

Hey, that works for me

I wil check out the web site - thanks. It looks very interesting!

NickiStone

I am a data/excel freak so I appreciate the how you have been breaking down all the data into usable forecasts. Also gotta love the charts! I may need to add some to my spreadsheets.

nuimproved

bump

dengarrett

So I changed the chart to use a seven day moving average as was suggested by PoleBoy. This shows that my results are MUCH closer than I thought AND the period of greatest deviation can be explained - it is when I was on vacation! I didn't eat that badly, but my exercise level was way down. This is exacly what I wanted to achieve - have the recorded track with the actual and be able to explain the deviations. After I see this pattern continue, I will then begin working on getting the actual and recorded match the planned - probably by changing the planned

afries0907

I see that you take a calcium supp. may i suggest that most people who take calcium also take a vit. d supp. Most calcium supp.s have vit d in them but some dont at the same time. the vit. d helps the calcium absorb into your system better:)

dengarrett

I see that you take a calcium supp. may i suggest that most people who take calcium also take a vit. d supp. Most calcium supp.s have vit d in them but some dont at the same time. the vit. d helps the calcium absorb into your system better:)

Thanks - actually I take Caltrate which is a Calcium & Vitamin D combination (recommended by my Dr.) But I didn't know it was to improve the absorption. Great to know!

givprayz

I thought I was a nerd keeping graph paper charts of daily weight with data about exercise, calories and sodium (as well as a few other pieces of TMI), but here I find out I'm barely a nerd at all! Sorry, I think it's cool.

I was going to suggest using a little less data, which would equate to the every 7 days approach. There are just too many daily variables. Over the last two weeks my weight has gone up as much as 4 pounds in 5 days, and down by as much as 2 pounds overnight. I just pass off all the increases as water weight, and the losses as real. My second chart only has 4 points per month, so the trend is almost 100% downward.

If you get motivated with all this, then by all means keep it up. If it gets to be a burden, drop it. I'm not trying to be negative. I think we need people willing to take on this kind of data collection. I'm impressed with your work, but I hope you don't make it about the data, and make it about a better life. Best of luck.

iamhealingmyself

Yes that is the math I used to get the actual calories burned. I agree that there are many factors that make it less than perfect (and I have learned from reading posts that women have more variability with TOM - see I even learned the acronym). But the main point is that we are planning and recording based on the belief that we ARE producing the actual results. And we ARE otherwise why do it? I think the real value in tracking like this is to ACCOUNT for the variables we cannot control. To be able to look at the deviations and say "oh that's when I overloaded on sodium" or "that was my TOM".

Saying it a different way - once we are able to predict and measure the effect our actions have on our weight then we can take responsibility for them and stop blaming ourselves for the things we have no control over.

Also - the effects of most items that are beyond our control last a short period of time. Over a longer period of planning, recording and measuring these should even out and we can see that we are able to control our weight.

These are all just my thoughts and observations so I welcome the comments and discussion.

When I get home this evening I will share the details on the spreadsheet.

I agree with your evaluation here. You're looking for the triggers that cause you to deviate from the plan so that you can better prepare for or anticipate them and not be confuzzled when you've suddenly gained 3 lbs and have been exercising so hard.

A lot of us are really unaware of what we're actually doing until we start logging, then the light bulb goes off. That leaves us with a general feeling of "now what" and a lack of direction. Cutting calories is one part of the overall equation but sometimes it's not enough (as in the case of a plateau where you actually need to increase and/or stagger calories to get the momentum back). This is an interesting aspect of what's really going on here. Maybe you should consider working with the site owner to incorporate it for everyone as part of the site. Seems like it could shed some really focused light on what we're doing.

Congratulations on your loss, your focus and your motivation!
Keep up the great work!

ETA after reading your research posts about population trials: If this was part of the site, you could gather that information seamlessly and for countless thousands if not millions of people. I would seriously consider at least discussing it with him (them).

dengarrett

I thought I was a nerd keeping graph paper charts of daily weight with data about exercise, calories and sodium (as well as a few other pieces of TMI), but here I find out I'm barely a nerd at all! Sorry, I think it's cool.

I was going to suggest using a little less data, which would equate to the every 7 days approach. There are just too many daily variables. Over the last two weeks my weight has gone up as much as 4 pounds in 5 days, and down by as much as 2 pounds overnight. I just pass off all the increases as water weight, and the losses as real. My second chart only has 4 points per month, so the trend is almost 100% downward.

If you get motivated with all this, then by all means keep it up. If it gets to be a burden, drop it. I'm not trying to be negative. I think we need people willing to take on this kind of data collection. I'm impressed with your work, but I hope you don't make it about the data, and make it about a better life. Best of luck.

Thanks - yeah I know I'm a nerd It is actually not a burden at all - all I have to do is add the additional row to the spreadsheet each day.

The seven day moving average idea did wonders for smoothing out the data (take a look at the graph at the end of the first page of this topic). And, it enabled me to see really where the deviation was (my vacation).

JeanneTops

Wow, at first I thought - way too much data. And then I remembered the spreadsheet I'VE been keeping for the last 15 months. Just weekly weight change in a chart and two graphs ;-) I wish now I had entered the calories as well. I just might go back and collect the data off of MFP although I think I'd have to do it manually.

Something my graph of 65 data points does show though is that weight loss does tend to fluctuate - there are even little bumps upward of 1 - 2 pounds plus some big drops of four pounds. I can only roughly align them with calorie consumption - holidays, vacations and so on - and the odd thing is that it doesn't always correlate. Sometimes I lose during vacation and gain 2 pounds on a stick-to-plan week. I wonder if maybe the body takes its own time to adjust - calorie loss this week, weight loss the next. Also, there seems to be a bounce factor - a big weight loss week will be followed by a small bounce up the next week. But the overall picture is a loss line with a fairly steady slope - its starting to flatten slightly now that I am at a much lower body weight.

So don't be concerned if recorded and actual don't always align even with a seven day rolling average. The more data you collect - meaning the more weight you lose! - the more you'll see that it's the long haul that is the most significant.

helenium

That looks great! I'm going to make one of my own now and try to analyse the plateau I just broke!

My only concern is that the BMR calculated by MFP is ONLY an estimate, and it may be more / less. Perhaps this will help me decide whether I'm more or less than my estimated BMR...

Edit: done... with moving average.



Interesting... my TOM came at the end of that period during the sudden rise. Impossible to say whether it was due to that though. Shows that I've got a discrepency between my predicted deficit and empirical deficit of about 2000 calories. Maybe that's due to come off soon ... it'll be better when I have a bit more data. This is fun!

dengarrett

BMR accuracy is one variable I have thought of also. Last night I started to play with the data a little to see what I would have to change to make the recorded (I like your term of predicted better) and actual to match more closely. That led me to wonder what are all the possible reasons for the deviations. Here are some I can think of:

Fat loss / gain
Muscle loss / gain
Water loss / gain
BMR being wrong
Daily activity burn being wrong
Exercise burn being wrong
Calorie consumption being wrong

I think those are the variables. Everything else is really just behavioral aspects related to them (for example, calorie consumption being wrong could be because all calories are not recorded or the food databae is wrong).

Thoughts?

hemlock2010

Just trying to wrap my head around the data, here . . .

So your moving average deficit is taken from your food diary (the # in the Remaining row at the end of the day--adding the last 7 days and dividing by 7) correct?

And your moving average actual is determined by multiplying pounds lost by 35000/lb (also 7-day moving average)?

So you should be able to tell whether MFP's guess for your BMR is accurate/close/or completely off-base by how well your actual weight loss tracks with MFP's data, correct? I'm asking b/c I've wondered about that, too, and this seems like a cool way to get an empirical answer.

That looks great! I'm going to make one of my own now and try to analyse the plateau I just broke!

My only concern is that the BMR calculated by MFP is ONLY an estimate, and it may be more / less. Perhaps this will help me decide whether I'm more or less than my estimated BMR...

Edit: done... with moving average.



Interesting... my TOM came at the end of that period during the sudden rise. Impossible to say whether it was due to that though. Shows that I've got a discrepency between my predicted deficit and empirical deficit of about 2000 calories. Maybe that's due to come off soon ... it'll be better when I have a bit more data. This is fun!

dengarrett

For me, the chart is plotting 7 day moving averages of three items:

Planned daily deficit (1000 per day)
Expected daily deficit based on the MFP diary
Actual daily deficit based on weight difference multiplied by 3500

Lots of imperfection in the actual buy I believe must be useful per previous discussions. What I hope to do now is explain those imperfections and account for them so that the predicted amounts are reliable indicators of the actual amounts. In effect, this proves the effectiveness of my "tuning knobs" which are the adjustments I make in diet, exercise and nutrition in general.

helenium

Just trying to wrap my head around the data, here . . .

So your moving average deficit is taken from your food diary (the # in the Remaining row at the end of the day--adding the last 7 days and dividing by 7) correct?

And your moving average actual is determined by multiplying pounds lost by 35000/lb (also 7-day moving average)?

So you should be able to tell whether MFP's guess for your BMR is accurate/close/or completely off-base by how well your actual weight loss tracks with MFP's data, correct? I'm asking b/c I've wondered about that, too, and this seems like a cool way to get an empirical answer.

All of this is right, except that the moving average is determined by multiplying by 3500/lb (not 35000) and yes it is also a moving day average.

This should be a good way to determine if MFP has your BMR right, but I think it is more effective over a period of 2 - 3 months with steady weight loss (hence my plateau throws it off). If you burn more than you think you should have, your BMR might be higher; if you burn less your BMR is lower.

hemlock2010

I set up the spreadsheet and back entered data to April 1. I don't have actual daily weights for that first week, so I just used my weekly weight. I'll post my graphs at the end of the month.

I want to track minutes of exercise (or calories burned? not sure) in cardio vs weights on a daily basis so that I can look at what happens to weight loss when I tweak those. I haven't added that to the spreadsheet yet--might do that later today.

dengarrett

I set up the spreadsheet and back entered data to April 1. I don't have actual daily weights for that first week, so I just used my weekly weight. I'll post my graphs at the end of the month.

I want to track minutes of exercise (or calories burned? not sure) in cardio vs weights on a daily basis so that I can look at what happens to weight loss when I tweak those. I haven't added that to the spreadsheet yet--might do that later today.

Yep - I started working on that today also - GMTA I want to do it because I back out my BMR calories during my period of exercise which is something I want to tweak. But... I gotta stop playing for now and get my car in for an oil change, get my hair cut, etc......

dengarrett

Ok, this is cool. Based on changes I know I made, I applied a few adjustments to the data at the point in time that I know I made them. For example, around the time of 3/24 I made two significant changes - 1) I began paying much closer attention to my nutrient targets by taking a vitamin supplement and lowering my sugar intake, and 2) increased the intensity of my exercise from 15-40 minute stints of Wii based exercise routines to hour long stints on my stationary recumbent bike.

So, I applied adjustments to the data to get the recorded and actual points to line up at about 3/24, and then dramatically increased the number of calories burned during exercise to the point that the lines track almost identically. And, interestingly enough, it appears that if I used MFP's calorie counts would be much more accurate than my HRM.

So, I am going to test my theory from here and begin to use MFP's calorie counts for 60 minute stationary bike workouts and see if the pattern continues. Note this is going to mean that the number of burned calories that I record is going to increase DRAMATICALLY from like 400 to 1600 per workout. If the patterns continue then I have to believe that is more accurate. If it doesn't, then I have to start looking at diet for the differences.

It sure is going to be an ego boost to see those kinds of numbers! :happy:

dengarrett

Woah! Wait a minute. Based on these assumptions, On days that I have worked out on the bike, I have burned twice the number of calories than I have eaten. Hard to imagine that is true, but the data is there. I lost about 8 pounds between 3/24 and now. That averages out to about 1800 calories per day burned. And if I look at my data with the adjustments made, it confirms those numbers.

Now, I just am not sure what to do. My instincts tell me to slow down the rate of loss a bit and eat more on the days I work out.

Gotta think about this some more.....

helenium

I think the close match between your recorded (/predicted) / actual is impressively accurate and it suggests that your BMR is pretty much spot on.

I don't know if you can make the conclusion that if you switch to MFP calorie counts and the trend continues, then the MFP calorie counts are better adjusted than your HRM. It might be compensating for some other factor being out, such as the ones you mentioned previously. Best look into other difference factors (including diet) instead.

One thing that would be worth explaining is the wavey patterns at the beginning of your graph where you get off to a slow start and accelerate beyond your predicted deficit. This might be due to muscle mass loss rather than fat loss, which would produce that kind of acceleration due to it being more dense than fat (and yet calculated as fat).

In the short term, water retention or dehydration that lasts > 5 days may produce differences of a few thousand calories that would show up on your graph despite the 7 day moving average.

The only other thing that springs to mind at the moment is monitoring changes in macronutrients, fibre and water consumption and relating them to short-term changes in weight, and that would be very in-depth analysis indeed.