Welcome to rlssm’s documentation

rlssm is a Python package for fitting reinforcement learning (RL) models, sequential sampling models (DDM, RDM, LBA, ALBA, and ARDM), and combinations of the two, using Bayesian parameter estimation.

Parameter estimation is done at an individual or hierarchical level using PyStan, the Python Interface to Stan. Stan performs Bayesian inference using the No-U-Turn sampler, a variant of Hamiltonian Monte Carlo.

Getting started:

• Installation
  • Dependencies
  • Conda environment (suggested)
• References
• Credits

How to:

• How to initialize a model
• How to fit a model
  • Non-learning example (non-hierarchical, simulated data)
  • Learning example (hierarchical, real data)
  • Diagnostics
  • Save the results
  • Re-load previously saved results
• How to inspect model fit results
  • Posteriors
  • Posterior predictives
  • Plot posterior predictives

Examples:

• Fit the DDM on individual data
• Fit the DDM on hierarchical data
• Parameter recovery of the DDM with starting point bias
• Parameter recovery of the hierarchical DDM with starting point bias
• Fit a RL model on individual data
• Fit a RL model on hierarchical data
• Fit the RLDDM on individual data
• Fit the LBA on individual data

API Reference:

• Model classes
• ModelResults class for RL models
• ModelResults class for DDMs (or RLDDMs)
• ModelResults class for race (or RL+race) models
• Simulate data with the DDM
• Simulate data with race models
• Simulate data with RL models, RLDDMs, and RL+race models

Index

• Index