search

Hussein-Mahfouz / drt-potential

3 stars 0 forks source link

Create activity-based model #19

Open Hussein-Mahfouz opened 6 months ago

Hussein-Mahfouz commented 6 months ago

Rough notes on steps involved:

Step 1: Generate synthetic population

I don't need to do this myself. I could use the output from any of: spenser | spc | david alvarez approach

Step 2: Add activity patterns to population

  1. explanation: each person in the synthetic population is given a daily schedule that shows how they spend their time. It includes trip chains, activity types, and mode of travel for each trip. At this point, we are yet to assign these activities to actual geographic locations
  2. method: statistical matching (Predhumeau and Manley 2023)
  3. data requirements: National Travel Survey or Time Use Survey
  4. code: ?

Step 3: Assign activities to geographic locations

  1. explanation: we take the trip chains assigned to each individual, and project them onto geographic locations. The assignment is done while considering constraints (specifically journey time distance between activity A and B)
  2. method: spatial sampling / space-time prisms (Hörl and Axhausen 2021)
  3. data requirements:
    • Activity locations: these can be queried from OSM. The ONS has a POI dataset for Britain. Arup also have a package (osmox) for this exact purpose
    • Individuals with activity chains (from step 2)
    • PT supply? Do we need a tt matrix between locations showing which OD pairs can be reached by which mode? Probably
  4. code:
    • sampling locations while accounting for journey time constraints: pam code
    • possibly useful code on discretionary locations - need to check how it is different to 1st approach / whether I will create a datset with non-mandatory activities

Step 4: Validation