autohrf {autohrf}

autohrf

Description

A function that automatically finds the parameters of model's that best match the underlying data.

Usage

autohrf(
  d,
  model_constraints,
  tr,
  roi_weights = NULL,
  allow_overlap = FALSE,
  population = 100,
  iter = 100,
  mutation_rate = 0.1,
  mutation_factor = 0.05,
  elitism = 0.1,
  hrf = "spm",
  t = 32,
  p_boynton = c(2.25, 1.25, 2),
  p_spm = c(6, 16, 1, 1, 6, 0),
  f = 100,
  cores = NULL,
  autohrf = NULL,
  verbose = TRUE
)

Arguments

d

A dataframe with the signal data: roi, t and y. ROI is the name of the region, t is the timestamp and y the value of the signal.

model_constraints

A list of model specifications to use for fitting. Each specification is represented as a data frame containing information about it (event, start_time, end_time, min_duration and max_duration).

tr

MRI's repetition time.

roi_weights

A data frame with ROI weights: roi, weight. ROI is the name of the region, weight a number that defines the importance of that roi, the default weight for a ROI is 1. If set to 2 for a particular ROI that ROI will be twice as important.

allow_overlap

Whether to allow overlap between events.

population

The size of the population in the genetic algorithm.

iter

Number of iterations in the genetic algorithm.

mutation_rate

The mutation rate in the genetic algorithm.

mutation_factor

The mutation factor in the genetic algorithm.

elitism

The degree of elitism (promote a percentage of the best solutions) in the genetic algorithm.

hrf

Method to use for HRF generation.

t

The t parameter for Boynton or SPM HRF generation.

p_boynton

Parameters for the Boynton's HRF.

p_spm

Parameters for the SPM HRF.

f

Upsampling factor.

cores

Number of cores to use for parallel processing. Set to the number of provided model constraints by default.

autohrf

Results of a previous autohrf run to continue.

verbose

Whether to print progress of the fitting process.

Value

A list containing model fits for each of the provided model specifications.

Examples

# prepare model specs
model3 <- data.frame(
  event        = c("encoding", "delay", "response"),
  start_time   = c(0,          2.65,     12.5),
  end_time     = c(3,          12.5,     16)
)

model4 <- data.frame(
  event        = c("fixation", "target", "delay", "response"),
  start_time   = c(0,          2.5,      2.65,    12.5),
  end_time     = c(2.5,        3,        12.5,    15.5)
)

model_constraints <- list(model3, model4)

# run autohrf
df <- flanker
autofit <- autohrf(df, model_constraints, tr = 2.5,
                   population = 2, iter = 2, cores = 1)