BTYD (version 2.4.3)

dc.compress.cbs: Compress Customer-by-Sufficient-Statistic (CBS) Matrix

Description

Combines all customers with the same combination of recency, frequency and length of calibration period in the customer-by-sufficient-statistic matrix, and adds a fourth column labelled "custs" (with the number of customers belonging in each row).

Usage

dc.compress.cbs(cbs, rounding = 3)

Arguments

cbs

calibration period CBS (customer by sufficient statistic). It must contain columns for frequency ("x"), recency ("t.x"), and total time observed ("T.cal"). Note that recency must be the time between the start of the calibration period and the customer's last transaction, not the time between the customer's last transaction and the end of the calibration period.

rounding

the function tries to ensure that there are similar customers by rounding the customer-by-sufficient-statistic matrix first. This parameter determines how many decimal places are left in the data. Negative numbers are allowed; see the documentation for round in the base package. As of the time of writing, that documentation states: "Rounding to a negative number of digits means rounding to a power of ten, so for example round(x, digits = -2) rounds to the nearest hundred."

Value

A customer-by-sufficient-statistic matrix with an additional column "custs", which contains the number of customers with each combination of recency, frequency and length of calibration period.

Details

This function is meant to be used to speed up log-likelihood and parameter estimation functions in the Pareto/NBD (pnbd) set of functions. How much faster those function run depends on how similar customers are. You can used compressed CBS matrices in BG/NBD estimation too, but there will be no speed gains there over using un-compressed CBS data.

This function only takes columns "x", "t.x", and "T.cal" into account. All other columns will be added together - for example, if you have a spend column, the output's spend column will contain the total amount spent by all customers with an identical recency, frequency, and time observed.

Examples

# NOT RUN {
# Create a sample customer-by-sufficient-statistic matrix:
set.seed(7)
x <- sample(1:4, 10, replace = TRUE)
t.x <- sample(1:4, 10, replace = TRUE)
T.cal <- rep(4, 10)
ave.spend <- sample(10:20, 10, replace = TRUE)
cbs <- cbind(x, t.x, T.cal, ave.spend)
cbs

# If cbs is printed, you would note that the following
# sets of rows have the same x, t.x and T.cal:
# (1, 6, 8); (3, 9)

dc.compress.cbs(cbs, 0)   # No rounding necessary

# Note that all additional columns (in this case, ave.spend)
# are aggregated by sum.
# }