Hi, Russ here.
Ever since I released this app, I've been slightly dissatisfied with the two smoothing algorithm choices. Each smoothing technique had its own weakness:
For this update, I rolled up my sleeves and researched new statistical techniques for smoothing real-time data. But none of them had all the characteristics I was looking for.
Then I thought: What if I created a new smoothing algorithm that smoothed a weight based not only on past weights, but also future weights? If I created a new algorithm that was allowed to make slight adjustments to the last couple of days of smoothed values as new data were recorded, could I create a new technique that captures the best parts of Exponential Smoothing and Double Exponential Smoothing? So I tried to mesh them, and low and behold, there's now a new technique that is the best of everything! I call it “Happy Scale Smoothing.”
For existing users, this new technique is an optional opt-in. You can find it in the settings under the “How To Calculate” heading.
Give the new “Happy Scale Smoothing” a shot! I think it’s the best option for weight smoothing available anywhere. Grab the update here.
- Russ