This PR updates cost_shifters.yaml with cost-shifter values from the first completed calibration run. Feel free to do a test simulation run.
Note: The calibrated cost-shifter values assume you are simulating with the changes in #10 and #11.
Calibration Target Data:
Observed proportion of real estate supply growth by city_id in the 2010 - 2015 period, for residential and non-residential separately
Calibration process:
Iteratively simulate the 2016 - 2018 period without scheduled development events
Compare simulated residential/non-residential supply growth proportions by city_id with the observed proportions from 2010 - 2015
Adjust cost-shifters to guide simulation in the direction of observed growth trends. Overshoot growth -> increase city cost. Undershoot growth -> decrease city cost
Terminate calibration once either correlation between simulated/observed exceeds .9 or once cost shifters reach excessively high values
Calibration goal:
Move relative spatial variation of simulated supply growth towards observed patterns
Proxy for unobserved costs and variables not accounted for by the proforma mechanics
The goal is NOT about getting the aggregate amount of 2015 - 2045 development right, so its possible that property value scaling may still be needed if too little aggregate supply growth
This PR updates cost_shifters.yaml with cost-shifter values from the first completed calibration run. Feel free to do a test simulation run.
Note: The calibrated cost-shifter values assume you are simulating with the changes in #10 and #11.
Calibration Target Data:
Calibration process:
Calibration goal: