Project Meeting 2022.03.29

Link to presentation - Shadow pricing.pptx

Agenda

Shadow pricing – constant added to each destination alternative in order to ‘doubly constrain’ the model. For example, the work location choice model is doubly constrained to match the origin constraints (workers) to destination constraints (employment).
It can be calculated in two ways
1. added as an alternate specific constant for each destination zone
2. multiplier on size term
Current approach in activity sim: the shadow prices are segmented by income group, and size terms are used as targets instead of employments as targets. This methodology doesn’t guarantee convergence to the total/distribution of employments in a zone.
Proposed methodology:
1. Calculate shadow prices as additive terms
2. use scaled total employment at different geographies as targets for work zone locations.
3. Introduce person-level weights when calculating modeled ratios.
4. Change iteration mechanisms

Workers who work more than one job are currently not modeled in activity sim. It was suggested to include as a separate GitHub issue and picked up later.
In the proposed shadow pricing algorithm, the employments are matched by total employment and not by employment in different categories.
Need metrics to measure the impacts of proposed shadow pricing.
Challenge to include shadow prices based on employment type industries. Industry type information is not included in the synthetic population. The focus of the algorithm is to make it as generic as possible that can be applied regardless of different employment types.
Need to do sensitivity tests to do measure the changes in trip length frequency distribution.

RSG will write all the proposed enhancements for shadow pricing and submit for further comments