Skip to content

SAPPHIRE is a lightweight, yet versatile simulation program that allows the integration of psychological theories, historical data, and heuristics to approximate human decision-making

License

Notifications You must be signed in to change notification settings

khnshn/SAPPHIRE

Repository files navigation

SAPPHIRE logo

SAPPHIRE: SimulAting ParticiPant beHavIoR during Experiments

SAPPHIRE version license

Instructions

Running the default simulation

import simulation as sim

director = sim.Director()
builder = sim.ConcreteBuilder()
director.builder = builder
director.build_default_simulation()
esm_sim = builder.simulation
x, m, u, b, r, c = esm_sim.simulate()

esm_sim.simulate() returns:

x: x-ticks array, m: memory accessibility array, u: urge (motivation) array, b: beeps array, r: responses array, and c: context array

the length of the arrays are equal

Creating a custom simulation

The parameters of the simulation can be customized by calling the setter methods of the builder. for example:

builder.set_days(100)
builder.set_random_beep(False)

Getting the parameter values of the simulation

print(esm_sim.get_attrib_value_list())

Output (values below correspond to the default builder):

[('_b', 0), ('_beep_power', 1), ('_beeps_a_day', 5), ('_context_aware_beep', True), ('_context_aware_threshold', 0.5), ('_context_change_steps', 2), ('_context_estimator_function', None), ('_context_reference', array([], dtype=float64)), ('_days', 25), ('_hours_in_day', 12), ('_max_c', 0.7), ('_min_c', 0.33), ('_override_inter_notification_steps', -1), ('_predefined_context_info', []), ('_predefined_times', []), ('_r', 0.2), ('_random_beep', False), ('_rate', -0.8), ('_salience_decay', 0.01), ('_salience_growth_ratio', 0.0015), ('_total_beeps', -1), ('_urge_decay', 0.005), ('_urge_max', 1), ('_use_prefedined_times_to_beep', True), ('decide', <bound method Simulation.decide of <simulation.Simulation object at 0x011D6BC8>>), ('est_context', <bound method Simulation.est_context of <simulation.Simulation object at 0x011D6BC8>>), ('exp_function', <bound method Simulation.exp_function of <simulation.Simulation object at 0x011D6BC8>>), ('get_attrib_value_list', <bound method Simulation.get_attrib_value_list of <simulation.Simulation object at 0x011D6BC8>>), ('is_in_fogg_signal_segment', <bound method Simulation.is_in_fogg_signal_segment of <simulation.Simulation object at 0x011D6BC8>>), ('linear_function', <bound method Simulation.linear_function of <simulation.Simulation object at 0x011D6BC8>>), ('random_with_change', <bound method Simulation.random_with_change of <simulation.Simulation object at 0x011D6BC8>>), ('send_beep', <bound method Simulation.send_beep of <simulation.Simulation object at 0x011D6BC8>>), ('simulate', <bound method Simulation.simulate of <simulation.Simulation object at 0x011D6BC8>>)]

Plotting the simulation output

import utils

fig, plt = utils.plot_sim(x, m, u, b, r, c)
plt.show()

An example output plot of the simulaiton in web env:


SAPPHIRE example (web)

Simulation parameters

Parameter Type Description Default
days (int; optional) the experiment period as the number of days 25
hours_in_day (int; optional) the total hours in a day that a beeps are allowed 12
beeps_a_day (int; optional) the (maximum) number of allowed beeps per day 5
context_change_steps (int; optional) the number of time steps the context remains unchanged. applies only when lenght of predefined_context_info is 0 or context_estimator_function is None 2
beep_power (float; optional) the maximum memory accessibility when a beep is received in percentage 1
urge_max (float; optional) the maximum possible motivation in percentage 1
r (float; optional) slope of the urge/motivation regeneration function 0.2
b (float; optional) intercept of the urge/motivation regeneration function 0
rate (float; optional) decay rate of the exponential function of memory accessibility -0.8
salience_decay (float; optional) the amount the beep_power drops each time a beep is received 0.01
urge_decay (float; optional) the amount the urge_max drops (increases) each time a beep is received at an inopportune (opportune) moment 0.005
randm_beep (bool; optional) to send beeps randomly or not. applies only when context_aware_beep is False False
min_c (float; optional) minimum probability of responding to beep given context. applies only when lenght of predefined_context_info is 0 or context_estimator_function is None 0.33
context_aware_beep (bool; optional) to send beeps according to the context or not True
predefined_times (list; optional) list of predefined beep times []
max_c (float; optional) maximum probability of responding to beep given context. applies only when lenght of predefined_context_info is 0 or context_estimator_function is None. 0.7
salience_growth_ratio (float; optional) the factor of salience_decay. used to regenerate salience over time 0.0015
predefined_context_info (list; optional) probability of responding to a beep given context at predefined times. must be mappable to predefined_times []
context_estimator_function (Any; optional) typically a trained ML model to estimate the probability of responding to a beep give a context None
context_reference (np.array; optional) list of predefined context information. must be mappable to predefined_times. must be an acceptable input to context_estimator_function np.array([])
context_aware_threshold (float; optional) the minimum acceptable probability of response based on a given context in order to send a beep 0.5
use_prefedined_times_to_beep (bool; optional) when True predefined_times is used True
override_inter_notification_steps (int; optional) when set to greater than -1 it will override the calculated inter notification time based on the experiment protocol -1
total_beeps (int; optional) when set to greater than -1 and when predefined_times exists it will determine the inter notification time so the total_beeps is achieved during the experiment period -1

Proposed architecture (SAPPHIRE builds the foundation for the Simulation Phase)


SAPPHIRE architecture

Publications

  • Alireza Khanshan, Pieter Van Gorp, and Panos Markopoulos. 2023. Simulating Participant Behavior in Experience Sampling Method Research. In Extended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems (CHI EA '23). Association for Computing Machinery, New York, NY, USA, Article 250, 1–7. https://doi.org/10.1145/3544549.3585586

About

SAPPHIRE is a lightweight, yet versatile simulation program that allows the integration of psychological theories, historical data, and heuristics to approximate human decision-making

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published