whatifbandit (version 0.3.0)

plot.mab: Plot Generic for mab objects

Description

Uses ggplot2::ggplot() to plot the results of a single Multi-Arm-Bandit trial. Provides options to select the type of plot, and to change how the plot looks. Objects created can be added to with + like any other ggplot plot, but arguments to change the underlying geom must be passed to the function initially.

Usage

# S3 method for mab
plot(x, type, level = 0.95, save = FALSE, path = NULL, ...)

Value

Minimal ggplot object, that can be customized and added to with + (to change scales, labels, legend, theme, etc.).

Arguments

x

A mab class object created by single_mab_simulation()

type

String; Type of plot requested; valid types are:

• arm: Shows Thompson sampling probabilities or UCB1 values over the trial period.

• assign: Shows cumulative assignment proportions over the trial period.

• estimate: Shows AIPW estimates for success probability with user specified normal confidence intervals based on their estimated variance.

level

Numeric value of length 1; indicates confidence interval Width (i.e 0.90, 0.95, 0.99). Defaults to 0.95.

save

Logical; Whether or not to save the plot to disk; FALSE by default.

path

String; File directory to save file if necessary.

...

Arguments to pass to ggplot2::geom_* function (e.g. color, linewidth, alpha, bins etc.).

Details

This function provides minimalist plots to quickly view the results of any Multi-Arm-Bandit trial, and has the ability to be customized through the ... inside the call and + afterwards. However, all the data necessary is provided in the output of single_mab_simulation() for extreme customization or professional plots, it is highly recommended to start completely from scratch and not use the generic.

The confidence intervals applied follow a standard normal distribution because it is assumed the AIPW estimators are asymptotically normal as shown in Hadad et al. (2021)

References

Hadad, Vitor, David A. Hirshberg, Ruohan Zhan, Stefan Wager, and Susan Athey. 2021. "Confidence Intervals for Policy Evaluation in Adaptive Experiments." Proceedings of the National Academy of Sciences of the United States of America 118 (15): e2014602118. tools:::Rd_expr_doi("10.1073/pnas.2014602118").

Examples

# Objects returned by `single_mab_simulation()` have a `mab` class.
# This class has a plot generic that has several minimal plots to examine
# the trial quickly

# Loading Data and running a quick simulation
data(tanf)
x <- single_mab_simulation(
  data = tanf,
  algorithm = "Thompson",
  assignment_method = "Batch",
  period_length = 600,
  whole_experiment = TRUE,
  perfect_assignment = TRUE,
  blocking = FALSE,
  prior_periods = "All",
  data_cols = c(
    condition_col = "condition",
    id_col = "ic_case_id",
    success_col = "success"
  )
)

# View best treatment arms over the simulation
y <- plot(x, type = "arm")
y
# Adding a new title
y + ggplot2::labs(title = "Your New Title")
# type = assign creates a similar plot, but shows probability of assignment instead

# Plotting Augmented Inverse Probability Estimates with confidence interval
# By default it provides 95% Normal Confidence Intervals but this can be adjusted
plot(x, type = "estimate")

# Adjusting height of internal geom* argument. (`geom_errorbarh()`)
plot(x, type = "estimate", height = 0.4)