Can someone point me to where the strength is computed?

[MatthewChatham]

From reading other issues and the FAQs I understand the app uses exponential smoothing. I'm a data nerd, so out of pure curiosity would anyone be able to point me to the bit of code that calculates the score?

[NellyWhads]

I think this is what you're looking for!

[iSoron]

This is also relevant. It's a very simple formula.

[naught101]

No much of a Java fan, but damn, that's some pretty code.

[Erudition]

     * The frequency of the habit is the number of repetitions divided by the
     * length of the interval. For example, a habit that should be repeated 3
     * times in 8 days has frequency 3.0 / 8.0 = 0.375.

Does this mean that completing the habit more than 3 times in an 8 day interval would exceed the 0.375 multiplier, thus artificially boosting your score? Since I expect the score is supposed to be about "how reliably do I satisfy the prescribed frequency" rather than "how frequently do I complete the habit", the score should not increase after the third completion in an 8-day period.

Update: Just tested with a new habit, indeed it does work as expected. Score goes up to 15% and stays there after 3 checks.
However, with any more than 3 checkmarks (4-8, starting 8 days ago) the graph shows nonzero data for the day before any checkmarks were ever added - a spike which drops down to zero on day 1. Hopefully just a weird caching issue from unchecking things.

[iSoron]

In some previous versions of the app, completing the habit more often than scheduled could make your score increase to 100% very quickly. In newer versions of the app, this is no longer possible. Performing a weekly habit twice per week doesn't make your score improve any faster than performing it once per week.

If you think the app is not calculating scores correctly, feel free to submit a detailed bug report. Screenshots are particularly helpful.