Weighted robust fit · per-staff daily throughput estimator
First row = header. Columns: date, staff, pending, processed, multiplier (0=closed, 0–1=partial, 1=open).
You can also reorder: set column indices below if your CSV differs.
Zero-indexed. Default assumes: date(0), staff(1), pending(2), processed(3), multiplier(4).
Drops top/bottom N% of implied per-staff rates before fitting. Handles Xmas dips & spikes.
Rows below this threshold are excluded (closed/near-closed days).
processed = C × staff × multiplier.
Outlier rows trimmed from each tail by implied rate before fitting.
Weight = (staff × multiplier)² — fully-staffed open days anchor the estimate.
Orange line = fitted capacity. Dots colored by multiplier (dark = fully open). Trimmed outliers shown as ×.