logit

What the dashboard is measuring

The dashboard is a rollup layer over logged workouts, not a hidden training-score engine. Most metrics come from simple date and exercise aggregations that are meant to stay legible: counts by day, counts by week, sums of logged volume, and maxima over stored exercise history.

That makes the numbers easier to trust. If a user wants to understand why a number changed, the explanation should usually trace back to a small set of logged workouts rather than to opaque scoring logic.

1. Workout counts and reporting windows

Workout counts are built from day-level aggregates keyed by each workout's performed date. Weekly and monthly summaries are then computed by summing those day counts inside the relevant date window.

c_t = \text{number of logged workouts on date } t

Interpretation. The overview screen is driven by literal counts of completed workout logs, grouped by their stored date.

The current week uses Monday as its starting boundary. The current month uses the normalized first day of the stored month. Weekly bars on the overview page are simply the current week's seven day buckets.

2. Month change

Month change is a relative comparison between workouts logged this month and workouts logged in the previous month. When the previous month has no workouts, logit avoids a divide-by-zero explosion and uses a simple product rule instead.

\mathrm{monthChange} = \begin{cases}\frac{W_{\mathrm{this}} - W_{\mathrm{prev}}}{W_{\mathrm{prev}}}\times 100 & W_{\mathrm{prev}} > 0\\100 & W_{\mathrm{prev}} = 0 \land W_{\mathrm{this}} > 0\\0 & \text{otherwise}\end{cases}

Interpretation. The percentage is only a true relative-change formula when the previous month is nonzero. Otherwise the display falls back to a bounded product convention.

3. Exercise summary records

Progress and overview views both depend on compressed exercise summaries. These are built by grouping normalized exercise history and then counting unique workout appearances, total sets, total reps, best logged load, and the most recent hit date.

\mathrm{sessionCount}_e = \left|\{\text{workoutId} : e \text{ appears in that workout}\}\right|

Interpretation. Session count is based on unique workouts, not on exercise rows. Best weight is the maximum logged weighted set for the grouped exercise history.

Metric	Implementation meaning	What it is not
Session count	Unique workouts containing that exercise	Number of times the exercise row appears
Best weight	Max logged load on any weighted set	An estimated 1RM or strength score
Total reps	Literal sum of logged reps for that exercise	A difficulty-adjusted workload
Last performed	Latest stored workout date for that exercise	The last time the user felt fresh on it

4. Progress-series metrics

The progress view compresses the last twelve Monday-start weeks into a small time series. For each week, it records the number of workouts and the sum of logged workout volume for that week.

\mathrm{weekDelta} = \mathrm{currentWeek} - \mathrm{previousWeek}

Interpretation. Week delta is an absolute count difference, not a percent. Average weekly sessions is the arithmetic mean across the twelve-week series.

Total lifted volume in this view is the lifetime sum of stored workout totals, later converted into the active display unit. It is not smoothed, normalized, or adjusted for effort.

5. Personal bests are literal logged bests

Personal best cards are selected by sorting exercises on their best logged load and taking the top five. That keeps the definition honest: the card shows the heaviest weight the user actually logged, with the most recent date associated with that exercise summary.

In other words, the dashboard does not currently elevate predicted capability above observed capability. The strongest-looking number on the screen is still grounded in a real logged set.