I have a small contention with the graphs. When I do not enter a value for a day, I honestly believe it should show that value as a 0. This will help portray an accurate graph (of how I am not working out), instead of only graphing the relative high points (when I do work out). I know there might be a problem of people complaining when a 0 is put in when they do not enter something, but that will just make the user more likely to actually enter in information. That's a good thing in my book. What do you think? If you don't find a value in the database, mark it as 0.

The one exception should be the weight graph. If a value isn't entered, I am under the impression it would be better to simply plot the previous days weight (the last recorded weight). Would this be possible?

At the very least, might be nice to be able to toggle this option--no entry = zero or no entry isn't shown as a marker point.

We could add an 'I did not work out today' button on the workouts section, to indicate that the day was, in fact, a zero, rather than just a failure to fill in, or we could do it as you describe, and give a 0 by default.

On the weight graph this would be possible, yes, but it makes more sense to show the trend between your last and current point - if you went from 220 to 210, it wasn't overnight, it was via 211-219,

We'll throw the idea around here a bit and see what we come up with - any other thoughts on the matter?

Thanks,

Maxwell

Good call on the weight thing. I can understand that. And actually, the "I did not work out today" button would be a most ideal situation. I actually like that better than having it auto set to zero.

I'd like to see more sensitive measures for the trend lines than connecting the first and last data points (in the case of weight measurements) or reporting the overall average (in the case of diet and calories burned working out).

I'd be especially interested in seeing an exponentially weighted moving average reported on the weight chart. The hacker's diet publishes a series of spreadsheets that demonstrate the use of this technique, and also give justifications for using it. Their info can be found here: URL.

Even just using simple linear regression for the weight trend would be better than the end-to-end line that is shown now.

Thanks so much for all of the great features on the site, can't wait to see what improvements are in store!

/au