1 of 6

HOPS

A Synthetic Population Forecasting Model

Draft

Roger Witte

22/4/26

OFFICIAL

OFFICIAL

2 of 6

CAVEAT

This is still ‘work in progress’: we are still finalizing the documentation and progressing the quality assurance ... Perhaps something will need to change as a result.

We also haven’t decided what will get published when and what licensing will apply. And finally, we now planning replace at least CTripEnd before making new NTEM trip end forecasts

OFFICIAL

OFFICIAL

3 of 6

The Context - NTEM

  • The National Trip End Model (NTEM) is a trip generation model of the potential demand for domestic surface travel by household residents in Great Britain.
  • The results are published and they are available for free download using the TEMPro application.
  • They include trips by purpose, time period, and mode at MSOA level (or Scottish equivalent)
    • Time periods are weekday AM peak, weekday interpeak, weekday PM peak, weekday off-peak, Saturday and Sunday
  • NTEM is commonly used in planning and transport planning applications such as the preparation of local plans, and transport appraisal.
  • The Department for Transport guidance, TAG, requires that all business cases submitted in support of funding in England should include modelling ‘consistent with NTEM over a sufficiently large area’
    • The aim is to ensure consistency of national assumptions between schemes in different places
    • Also used in Wales; inclusion of Scotland primarily to correctly represent Northern England
  • But the technological platform on which NTEM is built (Microsoft Access, Visual Basic and FORTRAN) is getting difficult to support

​

OFFICIAL

OFFICIAL

4 of 6

NTEM and HOPS

  • Logically NTEM proceeds as in the flow chart but the software only deals with aggregate totals
  • All the software needs replacement, but HOPS only does first half of the job
  • Output from HOPS is a 100% sample synthetic population of household residents in Great Britain
  • Includes a record for each synthetic person (linked to the household records)
  • Includes a record for each synthetic household (linked to the person records)
  • Can be aggregated within each MSOA to obtain same totals as we had previously
  • Can be used directly as input to activity scheduling demand model

Scenario Generator

Assemble Population into Households living in Dwellings

Forecast Jobs

NATCOP

Equip Households with cars

To be replaced by HOPS

CTripEnd

Forecasts Trip Ends

TEMPro

Forecasts Trip Ends

OFFICIAL

OFFICIAL

5 of 6

HOPS – Logical model

  • Constraint Data are the statistics that the model must fit in aggregate
    • Base constraint Data from census 2021 (Census 2022 in Scotland) and other official data sources
    • Growth factors, mostly estimated from published projections, applied to base data to get future year constraints
    • Flexibility about housing growth assumptions (so we can agree approach with MHCLG closer to when we are running forecasts)
  • If an area is overcrowded (high ratio of households to dwellings), households can move to neighbouring areas where there are vacant dwellings
  • NTS records used as micro data for Iterated Proportional Fitting
  • HOPS acts as a wrapper around the open source PopulationSim population synthesis engine originally developed in Oregon. HOPS chops the problem into pieces for parallel execution, organises the populationSim inputs and parameters for each piece, spawns the populationSim processes, and then combines the results.

OFFICIAL

OFFICIAL

6 of 6

Software Engineering aspects of HOPS

  • All source for execution and documentation of HOPS lives in a github site
    • Documentation source in markdown
  • Strict standards eg:
    • Accessibility : W3C
    • Python style : PEP8
  • Metadata in YAML format
    • Allows static type checking
    • Basis for ontology
  • Standards enforced with github actions
  • All project tasks/discussions handled through GitHub Issues
  • Test Driven Development
    • Tests written before python
    • Only ‘fake’ test data in github

Extensive use of discusussion in Github Issues, including recording code reviews and task assignments promoted ‘one team’ working between client and contractor.

OFFICIAL

OFFICIAL