bgcnbd.EstimateParameters: Parameter Estimation for the BG/CNBD-k model

Description

Estimates parameters for the BG/CNBD-k via Maximum Likelihood Estimation.

Usage

bgcnbd.EstimateParameters(cal.cbs, k = NULL, par.start = c(1, 3, 1, 3),
  max.param.value = 10000, trace = 0, dropout_at_zero = FALSE)

Arguments

cal.cbs

calibration period CBS. It must contain columns for frequency x, for recency t.x and total time observed T.cal. Optionally a column custs can be provided, which represents number of customers with a specific combination of frequency x, recency t.x and T.cal.

specified degree of regularity for Erlang-k distributed interpurchase times; needs to be integer-value; if this is not specified, then grid search from 1 to 12 is performed; this however requires column litt to be present in cal.cbs, which represents sum of logarithmic interpurchase times during calibration period;

par.start

initial BG/CNBD-k parameters - a vector with r, alpha, a and b in that order.

max.param.value

the upper bound on parameters

trace

print logging step every trace iteration

dropout_at_zero

Boolean; the mbg-methods are simple wrapper methods, which set this parameter to TRUE

Value

list of estimated parameters

References

Platzer Michael, and Thomas Reutterer (forthcoming)

Examples

Run this code

# NOT RUN {
cbs <- cdnow.sample()$cbs # load CDNow summary data
params <- bgcnbd.EstimateParameters(cbs)
# }