generateData: Generate Task Simulations

In jdtrat/dynaq: Tools to Simulate DynaQ Reinforcement Learning Algorithms

Description

Running this function is equivalent to running one agent on the two-stage Markov decision task.

Usage

generateData(
  numIter = 201,
  x = 10,
  alpha = 0.1,
  gamma = 0.9,
  epsilon = 0.1,
  tau = 0.08,
  softmax = TRUE
)

Arguments

numIter	The number of trials per session.
x	The amount of simulations to be done. This is the extent to which model-based activity occurs.
alpha	The learning rate alpha.
gamma	The temporal discounting factor gamma.
epsilon	The epsilon to be used in epsilon-greedy policy choices.
tau	The tau (temperature) to be used in softmax policy choices.
softmax	Logical: TRUE if softmax policy decisions should be used; FALSE if epsilon-greedy policy decisions should be used. By default, softmax is used.

Value

If x is greater than 0, so there are some model-based simulations, then this function returns a list with two elements containing real (element 1) and simulated (element 2) data. These can be passed into the processSimData function to concatenate into one dataframe. If x is equal to 0, so there are no model-based simulations, then this function returns a dataframe with real experience data.

jdtrat/dynaq documentation built on July 24, 2020, 7:18 a.m.