LinTSModel: Linear Thompson Sampling model.

In banditsCI: Bandit-Based Experiments and Policy Evaluation

Linear Thompson Sampling model.

Description

Creates a linear Thompson Sampling model for multi-armed bandit problems.

Usage

LinTSModel(
  K,
  p = NULL,
  floor_start,
  floor_decay,
  num_mc = 100,
  is_contextual = TRUE
)

Arguments

K	Integer. Number of arms. Must be a positive integer.
p	Integer. Dimension of the contextual vector, if is_contextual is set to TRUE. Otherwise, p is ignored. Must be a positive integer.
floor_start	Numeric. Specifies the initial value for the assignment probability floor. It ensures that at the start of the process, no assignment probability falls below this threshold. Must be a positive number.
floor_decay	Numeric. Decay rate of the floor. The floor decays with the number of observations in the experiment such that at each point in time, the applied floor is: floor_start/(s^{floor_decay}), where s is the starting index for a batched experiment, or the observation index for an online experiment. Must be a number between 0 and 1 (inclusive).
num_mc	Integer. Number of Monte Carlo simulations used to approximate the expected reward. Must be a positive integer. Default is 100.
is_contextual	Logical. Indicates whether the problem is contextual or not. Default is TRUE.

Value

A list containing the parameters of the LinTSModel.

Examples

model <- LinTSModel(K = 5, p = 3, floor_start = 1/5, floor_decay = 0.9, num_mc = 100,
                    is_contextual = TRUE)

banditsCI documentation built on April 12, 2025, 1:42 a.m.