JDWoodcock
commented
4 years ago We use the following data*: (what are they counting? Trips? People? Distance? Duration?) They are counting people at destination (thus conceptually I would consider this to be trips)
We augment motorcycle travel by adding person-less trips, that is, trips that don't belong to the synthetic population. Note that this means they don't contribute to the health burden.
We add trips totalling some distance according to how much fleet travel we expect to occur as a fraction of total motorcycle travel.
We use the following data*: (N.B. here, trips are counted, but we use the ratio in terms of distance. That means we have assumed trip distance does not vary between work and private use.)
Excluding NA, we use the values directly for cities where we have data:
For example, for Bogota, the fraction of total motorcycle travel from fleet is 244.8 / (244.8+310.08) if constant and ~ Beta (244.8 + 1 , 310.08 + 1 ) if uncertain (which is the posterior given a prior of Beta(1,1)).
For all Latin American cities for which we don't have direct data, we will use the Total, i.e. 937.91 / (937.91+822.6) if constant and ~ Beta (937.91/N , 822.6/N ) if uncertain, where we use the parameter N to increase uncertainty. E.g. N=80 puts 95% of the mass between 0.33 and 0.73, while the original cities are well bounded within [0.4,0.7].
*Rodríguez, D., Santana, M., & Pardo, C. (2015). La motocicleta en America Latina: caracterización de su uso e impactos en la movilidad en cinco ciudades de la región. (Despacio, Ed.). Bogotá: CAF. pages 149, 169, 190, 211, 233