Project implemented by Bartłomiej Chwiłkowski (github: chwilko)
- ADOPTION OF ELECTRIC VEHICLES – AN AGENT-BASED MODEL APPROACH
- Repository contents
## Model summary
Agent-based model simulating electric vehicle adoption for the Polish city of Wroclaw. The purpose of the study is to check how different government strategies influence the adoption of electric vehicles. In the model, environment interacts with government and car producers. Customers buy new car following the decision tree.
## Desctiption
The project is an impolementation of the model described in the master's thesis (At Wroclaw University of Science and Technology for Applied Mathematics). The thesis itself will be made available most likely when the polytechnic's right to first publication expires (6 months after submission).
# How to run model
First create virtual environment using venv:
python3.11 -m venv abm
source abm/bin/activate
python3 -m pip install -r requirements.txt
or using poetry (version 1.2.1)
poetry env use python3.11
poetry shell
poetry install
Next run
python3 -m model
Congrats, you just generated a sample plot. To generate the remaining plots perform:
bash generate_figs.sh
or open one of the ".ipynb" files from analyze/results/.
model
├── __init__.py
├── __main__.py
├── Cars.py
├── City.py
├── common.py
├── constants.py
├── Corporations.py
├── Customer.py
├── data.json
├── Government.py
├── initial_params.json
├── Prices.py
├── Society.py
├── Time.py
└── utils
├── __init__.py
├── common_params.py
├── get_trajectories.py
└── make_plot.py
The model is implemented according to the OOP paradigm.
Society is the main class. It is responsible for communicating with the other classes and iterating the algorithm. Class SocietyConstantsEnergyPrices is used by default. There are is SocietyVariableEnergyPrices too. Both of them inherit for Society.
Cars, as a customer field, are responsible for type (BEV, PHEV, CV), age and cost per km. Any car type has own class witch inherit for class Car (like abstract class).
The city creates the environment of the model. City is responsible for set initial public chargers, add new, remove old and count all nearby to checked Customer.
Corporation represent type of one model agent. Corporations are responsible for auto prices and their changes.
Customer represent type of one model agent. Customers are agents whose status will be counted and analyzed. Any Customer instance has indyvidual set of fields.
- procfile -- mean annual mileage
- home -- place in City
- car -- see class Car
The customer is also responsible for the mechanism for buying a car, and for choosing between two types.
Government represent type of one model agent. Government are responsible for introduce seted government strategy. Actually are implemented:
- GovernmentBuildChargingStation
- GovernmentProvidesSubsidies
- GovernmentMixedStrategy
- GovernmentNoSubsidies
- GovernmentCloseChargingStation
Any of them inherit for class AbstractGovernment (like abstract class). To check other strategies it is advisable to create a corresponding class inheriting from AbstractGovernment.
Price is responsible for fuel and energy prices. In the default society "SocietyConstantsEnergyPrices" (see Society) is used class ConstatntPrice and are constants prices. Howeover model is ready for implement function wich estimate future prices (class Prices). Both of them (ConstatntPrice, Prices) inherit for class Price.
Time is responsible for converting time from model time to real time. This is especially true for variable fuel and energy prices
In this folder are useful parts of the package not directly responsible for model implementations.
File common_params.py include json common_params with initial parameters used in thesis.
The get_trajectories.py file contains functions that allow multi-threaded generation of multiple model trajectories. Both for a fixed set of parameters and for a single parameter in different versions.
File make_plot.py get output with get_trajectories.py functions and create plots with matplotlib.
Not model parameters, but fixed values for customer profiles, and car parameters.
Initial distributions of profiles and car types.
The folder contains both data analysis and simulation files of various model trajectories.
In this folder I keep all collected data used to estimate model parameters. See corresponding README
Crowler used to download car data from otomoto.pl. Crowler is borrowed with small changes. See corresponding README If you want run crawler follow the instructions below:
First create virtual environment using venv:
python3.11 -m venv abm
source abm/bin/activate
python3 -m pip install -r requirements.txt
or using poetry (version 1.2.1)
poetry env use python3.11
poetry shell
poetry install
Next run scrapping using main
make scrap_om
or directly:
python3 -m otomoto-crawler
Folder for pictures generated by model. See section "How to run model"
Folder with tests. Tests use the pytest library. To run tests:
pytest
Makefile this is the control panel with endpoints:
make
Generation of graphs for trajectory and model with different parameters. See "How to run model".
make generate_figs
make lint
make run
See subsection "otomoto-crawler"
make scrap_om