Policy for the evaluation of policies with offline data through replay.

Details

The key assumption of the method is that that the original logging policy chose i.i.d. arms uniformly at random.

Take care: if the original logging policy does not change over trials, data may be used more efficiently via propensity scoring (Langford et al., 2008; Strehl et al., 2011) and related techniques like doubly robust estimation (Dudik et al., 2011).

Usage

  bandit <- OfflineReplayEvaluatorBandit(formula,
                                            data, k = NULL, d = NULL,
                                            unique = NULL, shared = NULL,
                                            randomize = TRUE, replacement = FALSE,
                                            jitter = FALSE)

Arguments

formula

formula (required). Format: y.context ~ z.choice | x1.context + x2.xontext + ... By default, adds an intercept to the context model. Exclude the intercept, by adding "0" or "-1" to the list of contextual features, as in: y.context ~ z.choice | x1.context + x2.xontext -1

data

data.table or data.frame; offline data source (required)

k

integer; number of arms (optional). Optionally used to reformat the formula defined x.context vector as a k x d matrix. When making use of such matrix formatted contexts, you need to define custom intercept(s) when and where needed in data.table or data.frame.

d

integer; number of contextual features (optional) Optionally used to reformat the formula defined x.context vector as a k x d matrix. When making use of such matrix formatted contexts, you need to define custom intercept(s) when and where needed in data.table or data.frame.

randomize

logical; randomize rows of data stream per simulation (optional, default: TRUE)

replacement

logical; sample with replacement (optional, default: FALSE)

replacement

logical; add jitter to contextual features (optional, default: FALSE)

unique

integer vector; index of disjoint features (optional)

shared

integer vector; index of shared features (optional)

Methods

new(formula, data, k = NULL, d = NULL, unique = NULL, shared = NULL, randomize = TRUE, replacement = TRUE, jitter = TRUE, arm_multiply = TRUE)

generates and instantializes a new OfflineReplayEvaluatorBandit instance.

get_context(t)

argument:

  • t: integer, time step t.

returns a named list containing the current d x k dimensional matrix context$X, the number of arms context$k and the number of features context$d.

get_reward(t, context, action)

arguments:

  • t: integer, time step t.

  • context: list, containing the current context$X (d x k context matrix), context$k (number of arms) and context$d (number of context features) (as set by bandit).

  • action: list, containing action$choice (as set by policy).

returns a named list containing reward$reward and, where computable, reward$optimal (used by "oracle" policies and to calculate regret).

post_initialization()

Randomize offline data by shuffling the offline data.table before the start of each individual simulation when self$randomize is TRUE (default)

References

Li, Lihong, Chu, Wei, Langford, John, and Wang, Xuanhui. Unbiased offline evaluation of contextual-bandit-based news article recommendation algorithms. In King, Irwin, Nejdl, Wolfgang, and Li, Hang (eds.), Proc. Web Search and Data Mining (WSDM), pp. 297–306. ACM, 2011. ISBN 978-1-4503-0493-1.

See also

Core contextual classes: Bandit, Policy, Simulator, Agent, History, Plot

Bandit subclass examples: BasicBernoulliBandit, ContextualLogitBandit, OfflineReplayEvaluatorBandit

Policy subclass examples: EpsilonGreedyPolicy, ContextualLinTSPolicy

Examples

if (FALSE) { url <- "http://d1ie9wlkzugsxr.cloudfront.net/data_irecsys_CARSKit/Movie_DePaulMovie/ratings.csv" data <- fread(url, stringsAsFactors=TRUE) # Convert data data <- contextual::one_hot(data, cols = c("Time","Location","Companion"), sparsifyNAs = TRUE) data[, itemid := as.numeric(itemid)] data[, rating := ifelse(rating <= 3, 0, 1)] # Set simulation parameters. simulations <- 10 # here, "simulations" represents the number of boostrap samples horizon <- nrow(data) # Initiate Replay bandit with 10 arms and 100 context dimensions log_S <- data formula <- formula("rating ~ itemid | Time_Weekday + Time_Weekend + Location_Cinema + Location_Home + Companion_Alone + Companion_Family + Companion_Partner") bandit <- OfflineReplayEvaluatorBandit$new(formula = formula, data = data) # Define agents. agents <- list(Agent$new(RandomPolicy$new(), bandit, "Random"), Agent$new(EpsilonGreedyPolicy$new(0.03), bandit, "EGreedy 0.05"), Agent$new(ThompsonSamplingPolicy$new(), bandit, "ThompsonSampling"), Agent$new(LinUCBDisjointOptimizedPolicy$new(0.37), bandit, "LinUCB 0.37")) # Initialize the simulation. simulation <- Simulator$new( agents = agents, simulations = simulations, horizon = horizon ) # Run the simulation. # Takes about 5 minutes: bootstrapbandit loops # for arms x horizon x simulations (times nr of agents). sim <- simulation$run() # plot the results plot(sim, type = "cumulative", regret = FALSE, rate = TRUE, legend_position = "topleft", ylim=c(0.48,0.87)) }