pnbd.PlotTrackingInc: Pareto/NBD Tracking Incremental Transactions Comparison

In BTYD: Implementing BTYD Models with the Log Sum Exp Patch

Description

Plots the actual and expected incremental total repeat transactions by all customers for the calibration and holdout periods, and returns this comparison in a matrix.

Usage

pnbd.PlotTrackingInc(
  params,
  T.cal,
  T.tot,
  actual.inc.tracking.data,
  n.periods.final = NA,
  xlab = "Week",
  ylab = "Transactions",
  xticklab = NULL,
  title = "Tracking Weekly Transactions"
)

Arguments

params	Pareto/NBD parameters - a vector with r, alpha, s, and beta, in that order. r and alpha are unobserved parameters for the NBD transaction process. s and beta are unobserved parameters for the Pareto (exponential gamma) dropout process.
T.cal	length of calibration period, or a vector of calibration period lengths.
T.tot	End of holdout period. Must be a single value, not a vector.
actual.inc.tracking.data	A vector containing the incremental number of repeat transactions made by customers for each period in the total time period (both calibration and holdout periods). See details.
n.periods.final	Number of time periods in the calibration and holdout periods. See details.
xlab	Descriptive label for the x axis.
ylab	Descriptive label for the y axis.
xticklab	Vector containing a label for each tick mark on the x axis.
title	Title placed on the top-center of the plot.

Details

actual.inc.tracking.data does not have to be in the same unit of time as the T.cal data. T.tot will automatically be divided into periods to match the length of actual.inc.tracking.data. See pnbd.ExpectedCumulativeTransactions.

The holdout period should immediately follow the calibration period. This function assume that all customers' calibration periods end on the same date, rather than starting on the same date (thus customers' birth periods are determined using max(T.cal) - T.cal rather than assuming that it is 0).

Value

Matrix containing actual and expected incremental repeat transactions.

Examples

data(cdnowSummary)
cal.cbs <- cdnowSummary$cbs
# cal.cbs already has column names required by method

# Cumulative repeat transactions made by all customers across calibration
# and holdout periods
cu.tracking <- cdnowSummary$cu.tracking
# make the tracking data incremental
inc.tracking <- dc.CumulativeToIncremental(cu.tracking)

# parameters estimated using pnbd.EstimateParameters
est.params <- cdnowSummary$est.params

# All parameters are in weeks; the calibration period lasted 39
# weeks and the holdout period another 39.
pnbd.PlotTrackingInc(params = est.params, 
                     T.cal = cal.cbs[,"T.cal"], 
                     T.tot = 78, 
                     actual.inc.tracking.data = inc.tracking)

BTYD documentation built on Nov. 18, 2021, 1:10 a.m.