Show HN: Happy Scale – Use moving averages for accurate weight loss predictions(happyscale.com)
happyscale.com
Show HN: Happy Scale – Use moving averages for accurate weight loss predictions
http://happyscale.com/
26 comments
Thanks for the feedback Kevin!
Oh man, can't believe that copy issue (triple-line instead of plus) has been in the first run experience for over a month and I didn't realize it... thanks for pointing it out!
Great insight about not hiding the feature list behind that button... hadn't thought about that since as a dev I know that button can never charge me w/out another prompt from Apple. Dev blindness.
Also, what an interesting (and new to me) idea about trying a conversion during onboarding. Apple would indeed allow it. Great opportunity for an A/B test.
Yes, defaulting to actual weights is something I've wrestled with. The fact is, while I knew I could play up the moving average angle with this audience, many of my customers got confused about how moving averages worked and why they were seeing weights they didn't type in on their second day of usage. Also, moving average predictions are pretty inaccurate until there's a week's worth of data, so it's really hard to have a great day 2 experience using just the moving averages. In my experience, defaulting to simple weight tracking for the numbers I display and letting power users switch to moving averages seems to have worked best.
Kevin- This is some of the most thought-provoking feedback I've gotten. You clearly put a good amount of your time into the review and the feedback, and I really appreciate how you gave actionable advice. Thanks for your help!
Oh man, can't believe that copy issue (triple-line instead of plus) has been in the first run experience for over a month and I didn't realize it... thanks for pointing it out!
Great insight about not hiding the feature list behind that button... hadn't thought about that since as a dev I know that button can never charge me w/out another prompt from Apple. Dev blindness.
Also, what an interesting (and new to me) idea about trying a conversion during onboarding. Apple would indeed allow it. Great opportunity for an A/B test.
Yes, defaulting to actual weights is something I've wrestled with. The fact is, while I knew I could play up the moving average angle with this audience, many of my customers got confused about how moving averages worked and why they were seeing weights they didn't type in on their second day of usage. Also, moving average predictions are pretty inaccurate until there's a week's worth of data, so it's really hard to have a great day 2 experience using just the moving averages. In my experience, defaulting to simple weight tracking for the numbers I display and letting power users switch to moving averages seems to have worked best.
Kevin- This is some of the most thought-provoking feedback I've gotten. You clearly put a good amount of your time into the review and the feedback, and I really appreciate how you gave actionable advice. Thanks for your help!
I think that is an error in the content, it says hit the triple line button in the upper right. Well one half of that is wrong, so if the + button did allow you to add a weight, then it meant to say plus instead of triple line.
Exactly. I changed this copy in the last build when I added HealthKit, so I'm hoping this slip up is at least limited in how long it's been there.
Hi HN!
Happy Scale helps people who are losing weight to redefine their relationship with the scale. By using their daily weigh-in as a single data point in a moving average instead of a source of truth, you can see daily weight loss progress in the app even when your scale weight plateaus for a few days. Over time, you become desensitized to the horror of a bad weigh-in and can avoid discouragement during the tender moments of building your new lifestyle habits.
I created this iOS app because I would be devastated in the first few days of a diet when I hopped on the scale and saw a number that was higher than my starting weight. I went on to learn that there are many factors that can cause fluctuations of a few pounds in our weight from day to day (glycogen stores, water retention, food in the belly, etc.), and using a moving average is a convenient way to neutralize these factors.
I've been working on this as a nights and weekends project for a bit over 4 years. I have some big ideas for where I'd like to take it in the future, but I'd love to get your feedback on where it is today!
Happy Scale helps people who are losing weight to redefine their relationship with the scale. By using their daily weigh-in as a single data point in a moving average instead of a source of truth, you can see daily weight loss progress in the app even when your scale weight plateaus for a few days. Over time, you become desensitized to the horror of a bad weigh-in and can avoid discouragement during the tender moments of building your new lifestyle habits.
I created this iOS app because I would be devastated in the first few days of a diet when I hopped on the scale and saw a number that was higher than my starting weight. I went on to learn that there are many factors that can cause fluctuations of a few pounds in our weight from day to day (glycogen stores, water retention, food in the belly, etc.), and using a moving average is a convenient way to neutralize these factors.
I've been working on this as a nights and weekends project for a bit over 4 years. I have some big ideas for where I'd like to take it in the future, but I'd love to get your feedback on where it is today!
Import function seems to be broken on iOS 9.0 beta 13A4325c:
http://imgur.com/cHs5G1t
http://imgur.com/cHs5G1t
Thanks for letting me know! If you'd like me to let you know when it's fixed, email me at [email protected].
Looks cool. Any android plans?
I'd love to, but I haven't had the time to port it. I've been putting all my energy into improvements on the iOS side.
Nothing against Android! It's just a huge investment to port it.
For what its worth take a look at https://medium.com/@periscope/periscope-by-the-numbers-6b23d... and guess when they released their android version. Notice anything?
For Android users, I've been using Libra for several years now and enjoy that I can customize the moving average.
http://www.appbrain.com/app/libra-weight-manager/net.cachapa...
https://play.google.com/store/apps/details?id=net.cachapa.li...
http://www.appbrain.com/app/libra-weight-manager/net.cachapa...
https://play.google.com/store/apps/details?id=net.cachapa.li...
I've never used this app first-hand, but it's the one I recommend to users when they ask about an Android version. I didn't know they let you customize the moving average, so I'll have to check that out.
If you wouldn't mind sharing, how do they let you customize it, and what do you enjoy about that?
If you wouldn't mind sharing, how do they let you customize it, and what do you enjoy about that?
Under 'Preferences'[1] there is an 'Advanced Preferences'. In Advanced Preferences, there are two relevant settings: Smoothing Days and Forecast days. The differences aren't clearly delineated, so when I first started using Libra (several years ago), I ended up experimenting for several weeks.
The explanation for smoothing days is that it "[controls] how strongly each weight value affects the trend". It takes an positive integer value or 0. The recommended value is 7, but I ended up with 14. With higher values, both the forecast line and average line tend to be less dependent on recent weights and average out over a longer period.
The forecast days "[control] how many days to use for the forecast calculation. It doesn't specify what the acceptable range is, simply that the minimum value is 2 and the recommended value is 7. I ended up with a 14 here also. This value only impacts the forecast line, not the average line.
[1] On my phone, reached via Android's hamburger physical button. [2] Reached via a button that looks like the ai-settings icon from the http://www.androidicons.com/ set.
The explanation for smoothing days is that it "[controls] how strongly each weight value affects the trend". It takes an positive integer value or 0. The recommended value is 7, but I ended up with 14. With higher values, both the forecast line and average line tend to be less dependent on recent weights and average out over a longer period.
The forecast days "[control] how many days to use for the forecast calculation. It doesn't specify what the acceptable range is, simply that the minimum value is 2 and the recommended value is 7. I ended up with a 14 here also. This value only impacts the forecast line, not the average line.
[1] On my phone, reached via Android's hamburger physical button. [2] Reached via a button that looks like the ai-settings icon from the http://www.androidicons.com/ set.
So you liked that you could customize it to be more responsive to recent weight changes... interesting.
I do also have an advanced settings section for the forecast calculation, but it's good to know there's interest in controlling the MV factors too. Thanks for the feedback!
I do also have an advanced settings section for the forecast calculation, but it's good to know there's interest in controlling the MV factors too. Thanks for the feedback!
From my experience, using a moving average works really well to track weight. I am repeatedly amazed that more apps don't have something like this considering Hacker's Diet[1] was doing it over a decade ago.
[1] http://www.fourmilab.ch/hackdiet/online/hdo.html
[1] http://www.fourmilab.ch/hackdiet/online/hdo.html
Very nice app! Thanks for putting decent HealthKit toggles in the app itself. I'm quite choosy about which apps can read HK data and which can write, and many of them are not good at making it obvious if they can cope with that!
Thanks! It's always a balance between keeping it simple and giving users control. I'm glad to hear from the segment that appreciates that control.
Looks like a nicer UI than the withings app. Any thoughts about adding bf%? My #1 feature request for an app like this would be to see trend lines for LBM kg and Total Fat kg on one chart.
I actually really like the Withings UI, so I take that as high praise.
Yes, Body Fat % is a pretty important thing on my list. I don't have it scheduled for the next release, but I do hope to get it into one of the next couple of releases.
Yes, Body Fat % is a pretty important thing on my list. I don't have it scheduled for the next release, but I do hope to get it into one of the next couple of releases.
Just to add a bit more feedback here, the part that withings fails me on is not exposing Fat and LBM in weight terms. If my goal is to maintain LBM and see Fat % go down over several weeks it makes it difficult to determine if that last 2% of bf% drop was too fast (read: too much LBM loss) without actually exporting the data and calculating things by hand. They used to make these variables available in their old web ui, but they aren't available in the app. It feels a bit dumbed down.
They also allow you to set a weight goal, but not a bf% goal, which is so confusing when their marquee product includes a body fat analyzer. For a lot of people who buy these scales to help gauge recomposition efforts the goal over time is to keep their weight the same but see the bf% go down.
They also allow you to set a weight goal, but not a bf% goal, which is so confusing when their marquee product includes a body fat analyzer. For a lot of people who buy these scales to help gauge recomposition efforts the goal over time is to keep their weight the same but see the bf% go down.
What's the 'advanced math' used for the moving average? Some form of convolutional filter?
The simple answer is an exponential moving average, but there's some additional stuff to compensate for the lag associated with using a moving average on a downward-trending or upward-trending dataset.
http://cl.ly/image/1d3q2Q3S0029
Why would I click on the hamburger menu icon to add a weight? Seems like I should hit the + button on the top right? In fact, it seems incorrect since it did what I expected, which was slide out the navigation. The + button, did ask me to add weight.
One thing I wanted (and maybe you have it?) was the ability to set a regular notification to remind me to weigh myself. Also, for the deluxe edition what are the other features besides removing the ads? I couldn't find what those would be. It showed up a nice list of them when I hit the button, but know that most people are going to be hesitant to do so. I'm not sure if Apple would allow it, but if you could integrate the question to upgrade into your onboard slideshow, I'd bet you'd get higher conversion. The way it is, it's pretty buried in there and this is the kind of app that people won't upgrade until they have several days worth of data to see if it's worth it. Chances are your app is going to tell them that they're bad at losing weight. They're going to be the most motivated and hopeful right when they downloaded it. Hope is a good place for conversion in my opinion.
Also, for the My Weight Goals, I didn't understand why it defaulted to Track Using : Actual Weights. Isn't the point to have it track using Moving Avg Weights? I just wasn't sure why you defaulted to this.
Thanks for sharing!