A Bayesian Factor Model has raw data and draws from the posterior distribution. This is represented by an R6 Class.
Code and theory based on Pinkney 2021.
public methods:
initialize() initializes the variables and model parameters
fit() fits the stan model and returns a fit object
updateWidth updates the width of the credible interval
placeboPlot generates a counterfactual placebo plot
effectPlot returns a plot of the treatment effect over time
summarizeLiftreturns descriptive statistics of the lift estimate
biasDraws returns a plot of the relative bias in a LFM
liftDraws returns a plot of the posterior lift distribution
liftBias returns a plot of the relative bias given a lift offset
vizdraws object with the relative bias with offset.
timeTilesggplot2 object that shows when the intervention happened.
plotDatatibble with the observed outcome and the counterfactual data.
interventionTimereturns the intervention time period.
syntheticggplot2 object that shows the observed and counterfactual outcomes over time.
new()Create a new bayesianFactor object.
bayesianFactor$new(
data,
time,
id,
treated,
outcome,
ci_width = 0.75,
covariates
)dataLong data.frame object with fields outcome, time, id, and treatment indicator.
timeName of the variable in the data frame that
idName of the variable in the data frame that identifies the units (e.g. country, region etc).
treatedName of the variable in the data frame that contains the treatment assignment of the intervention.
outcomeName of the outcome variable.
ci_widthCredible interval's width. This number is in the (0,1) interval.
covariatesDataframe with a column for id and the other columns Defaults to NULL if no covariates should be included in the model.
params described in the data structure section of the documentation of the R6 class at the top of the file.
A new bayesianFactor object.
fit()Fit Stan model.
bayesianFactor$fit(L = 8, ...)LNumber of factors.
...other arguments passed to rstan::sampling().
updateWidth()Update the width of the credible interval.
bayesianFactor$updateWidth(ci_width = 0.75)ci_widthNew width for the credible interval. This number should be in the (0,1) interval.
summarizeLift()summarizeLift returns descriptive statistics of the lift estimate.
bayesianFactor$summarizeLift()
effectPlot()effectPlot returns ggplot2 object that shows the effect of the intervention over time.
bayesianFactor$effectPlot()
liftDraws()Plots lift.
bayesianFactor$liftDraws(from, to, ...)fromFirst period to consider when calculating lift. If infinite, set to the time of the intervention.
toLast period to consider when calculating lift. If infinite, set to the last period.
...other arguments passed to vizdraws::vizdraws().
vizdraws object with the posterior distribution of the lift.
liftBias()Plot bias magnitude in terms of lift for period (firstT, lastT)
bayesianFactor$liftBias(firstT, lastT, offset, ...)firstTStart of the time period to compute relative bias over. Must be after the intervention.
lastTEnd of the time period to compute relative bias over. Must be after the intervention. over. They must be after the intervention.
offsetTarget lift %.
...other arguments passed to vizdraws::vizdraws().
biasDraws()Plots relative upper bias / tau for a time period (firstT, lastT).
bayesianFactor$biasDraws(small_bias = 0.3, firstT, lastT)small_biasThreshold value for considering the bias "small".
firstT, lastTTime periods to compute relative bias over, they must after the intervention.
vizdraw object with the posterior distribution of relative bias. Bias is scaled by the time periods.
clone()The objects of this class are cloneable with this method.
bayesianFactor$clone(deep = FALSE)deepWhether to make a deep clone.