chart-scaling-law
A four-state log-log scaling plot that fits L = E + A·C^(−α) to the runs it is given — one power law per model family, with the exponent, the irreducible loss and R² printed, the stretch past the largest run hatched and projected, and compute shares apportioned so every filter state totals exactly 100.
chart-scaling-law
A four-state log-log scaling plot that fits L = E + A·C^(−α) to the runs it is given — one power law per model family, with the exponent, the irreducible loss and R² printed, the stretch past the largest run hatched and projected, and compute shares apportioned so every filter state totals exactly 100.