Software and code

For statistical/computational modelling, we mainly use R. For experiments, we mainly use Python and html/javascript. R packages are hosted on CRAN and R-Forge, but we have recently started to use github.

DepmixS4

An R package to estimate mixture and hidden Markov models

DepmixS4 is written and maintained by Ingmar Visser and Maarten Speekenbrink and based on Ingmar’s earlier depmix package. DepmixS4 has an object-oriented design, providing a flexible interface to fit dependent mixture models with different observation models. Estimation is done by Expectation-Maximisation or numerical optimization of the likelihood using Rsolnp. The current stable release of depmixS4 can be found on CRAN. The latest (beta) version can be found on R-Forge site.

dlrm

An R package to estimate dynamic linear regression models

The Dynamic Lens Model (DLM, Speekenbrink & Shanks, 2010) is used to estimate cue utilization in MCPL tasks. Essentially, the model is a dynamic linear regression model (dlrm). Estimation relies on the Kalman filter/smoother and Expectation-Maximisation. The R package “dlrm” implements this and is hosted on R-Forge. It can be downloaded from this link. To install this package directly within R type:

install.packages("dlrm", repos="http://R-Forge.R-project.org")

At the moment, the documentation is minimal but we hope to correct this in the near future.