Skip to content

Latest commit

 

History

History
119 lines (90 loc) · 3.16 KB

journal.md

File metadata and controls

119 lines (90 loc) · 3.16 KB

How to run

python src/main.py -r 100 -js ex.json

tensorboard --logdir logs/2x256-1579515056

Resources

Runs

  • muda numero de pessoas para ver traffic:
    • 100 check
    • 300 check
    • 500 check
  • muda os custos
    • menos para autocarro check
    • mais para carro(portagem) +1 check
  • mudar o tempo
    • autocarro check
  • mudar conforto
    • autocarro check
  • mais pessoas a partilhar

3000 episodes

  • run 1

    • 100 pessoas
    • tudo igual
  • run 2

    • 300 pessoas
    • tudo igual
  • run 3

    • 500 pessoas
    • tudo igual
  • run 4

    • 1 custo autocarro
  • run 5

    • custo carro + 2 custo shared + 2/3
  • run 6

    • mudar tempo do autocarro + ou - dependendo se pessoas gostam do autocarro
  • run 7

    • mudar conforto do autocarro + ou - dependendo se pessoas gostam do autocarro

To Do

  • Add people that choose a mean of transport
  • Write stats to a file
  • Think about others possible parsers, maybe for population
  • Think about attributes the population will have - relationships
  • Utility function - paper
  • Reinforcement learning
  • Stats about occupancy of edges, type of actors, number people chose each service, utility
  • Bus event 10 in 10 minutes?
  • Incentives - think about how Cat can introduce them later in an easy breazy way
  • Utility factors- how to account for sociability? For now we are going to only have the awareness of the impact of the chosen service
  • Actor occupation chart uses only last few runs

Done

  • Make different actors - bus,car,shared car
  • Make edges only let certain actors traverse them
  • Create routes for actors according to the edges specifications
  • Parser for graph

Things to have until 15/12

  • Make different actors - bus,car,shared car
  • Make edges only let certain actors traverse them
  • Create routes for actors according to the edges specifications
  • Added Users
  • Utility function
  • Reinforcement Learning
  • Users can only choose private car if they have a car
  • Utility function can't be divided by 0

Might have

  • Gif showing the occupation of the edges
  • Gif showing people learning

Probs not

  • Each service has one color

09-12-2019

  • In the process of changing actors_flow to be a dictionary for all services (actors)

10-12-2019

  • Charts now show statistics for different actors - actor chart and edge chart

11-12-2019

  • Graph is created based on a json file (no longer hardcoded)

12-12-2019

  • Added Users
  • Users choose (randomly) a transportation method
  • Users have the "result" of the commute (time, cost, awareness)

13-12-2019

  • Added Reinforcement learning

14-12-2019

  • Tensorboard
  • Tried different personality types - need to work on that

14-04-2020

  • Finaly made a commit with all the changes as of yet
  • Users have more information from clusters, factors, courses and grades
  • Writing user info to a file at the end of the main (need to think about a new place)

18-02-2020

  • Users now have salary and budget
  • Fixed willingness to pay so that it now has to do with the salary of the user