Agent-based consumer population model for research tasks,
forecasting and working out scenarios

The model is intended to study the influence of external circumstances, advertising recommendations and management decisions on the consumer behavior of agents individually, various social groups and the population as a whole, as well as for training and testing recommender systems.

The development includes a consumer model with customizable individual characteristics, a model of the external environment that allows for setting scenarios and includes recommendations from the system under study, a set of examples of creating a population and implementing scenario modeling.

The input is real world data or a given scenario of crisis events or management impacts;

The output is simulated transactions of individual agents and their aggregations in the area of consumer interest, in groups of agents and in the population as a whole, allowing one to make forecasts of consumer behavior under various variants of external factors or to judge the effectiveness of recommendations or management decisions.

Content

simulation.ipynb - Jupyter notebook with demo example.

data - a folder with an example of real-world data on the environment for similation.

src/simulator.py - python procedures for modelling of a customer behavoiur, creation of population and transactional activity simulating

src/environ.py - example of environmental scenario realization

Process flow

Desk-top integration

Under construction...