different language between tool builder-modeler, policy maker-public pairs

[hyunjimoon]

this is a summary thread that collects every discussion on estimation problem from prarameterization. different language between tool builder-modeler, policy maker-public pairs is the cause of some parameterization. E.g. in defense spare part industry, modelers and policy makers communicate with time coordinate (e.g.MTBF, repair time), but in mathematical sense, it's better to use rate and adj_frac coordinate.

One example was reported discrepancy between Battle Field and Sum of Battle Fields which can be resolved (based on math on the right) after changing repair time from ("1Repair Time"+1/"2Repair Time")/2 to 1/((1/"1Repair Time"+1/"2Repair Time")/2) (1/(1/"1Repair Time"+1/"2Repair Time")/2 doesn't work for vensim!) Repair Estimation2.mdl.zip In the plot, red, green, gray are overlaped. The value for sum of battlefields is higher when using arithmetic mean compared to harmonic mean -- this may be explained by AM-GM-HM inequality (visual)

• time is better in elicitation from modeler, but adj_frac, rate is better for tool builder for estimation
• scale, prevent outlier, remove process noise for speedup #124
• Does use of flow measurements actually solve the autocorrelation problem? #127