quickpsy is an R package developed by Daniel Linares and Joan López-Moliner to quickly fit and plot psychometric functions for multiple conditions. It makes an extensive use of Hadley Wickham's packages ggplot2 and dplyr.
To understand the fundamentals of fitting psychometric functions in R, we recommend the book Modeling Psychophysical Data in R.
Features
-
Fits and plots multiple conditions with minimal coding.
-
Exploits the computational speed of dplyr.
-
The user does not need to introduce initial parameters.
-
Calculates parametric and non-parametric bootstrap confidence intervals.
-
Guess and lapses can be fixed or free as parameters.
-
Fits cumulative normal, logistic, weibull functions or any function defined by the user.
-
Facilitates the reading of several data files.
-
The optimization can call DEoptim which uses differential evolution optimization.
Install
Download and install R (we also recommend Rstudio).
In R, install the following packages: boot, DEoptim, dplyr, ggplot2, tidyr and devtools.
install.packages('boot')
install.packages('DEoptim')
install.packages('tidyr')
install.packages('devtools')
Install quickpsy from github (will also install dplyr and ggplot2) and load the package
install_github('danilinares/quickpsy')
Example
library(quickpsy)
library(MPDiR) # contains the Vernier data; use ?Venier for the reference
fit <- quickpsy(Vernier, Phaseshift, NumUpward, N,
grouping = .(Direction, WaveForm, TempFreq))
plotcurves(fit)
plotpar(fit)
plotthresholds(fit)
Help
To obtain information and examples for specific functions use ?
?plotcurves
For further information visit www.dlinares.org/quickpsy.html
Other R packages
-
psyphy: among other things, it provides links functions to fit psychometric functions using an approach based on generalized linear models.
-
modelfree: fits psychometric functions using a non-parametric approach.