Based on Croston's (1972) method for intermittent demand forecasting, also described in Shenstone and Hyndman (2005). Croston's method involves using simple exponential smoothing (SES) on the non-zero elements of the time series and a separate application of SES to the times between non-zero elements of the time series.
CROSTON(
formula,
opt_crit = c("mse", "mae"),
type = c("croston", "sba", "sbj"),
...
)A model specification.
Model specification (see "Specials" section).
The optimisation criterion used to optimise the parameters.
Which variant of Croston's method to use. Defaults to "croston" for
Croston's method, but can also be set to "sba" for the Syntetos-Boylan
approximation, and "sbj" for the Shale-Boylan-Johnston method.
Not used.
The demand special specifies parameters for the demand SES application.
demand(initial = NULL, param = NULL, param_range = c(0, 1))
initial | The initial value for the demand application of SES. |
param | The smoothing parameter for the demand application of SES. |
param_range | If param = NULL, the range of values over which to search for the smoothing parameter. |
The interval special specifies parameters for the interval SES application.
interval(initial = NULL, param = NULL, param_range = c(0, 1))
initial | The initial value for the interval application of SES. |
param | The smoothing parameter for the interval application of SES. |
param_range | If param = NULL, the range of values over which to search for the smoothing parameter. |
Note that forecast distributions are not computed as Croston's method has no underlying stochastic model. In a later update, we plan to support distributions via the equivalent stochastic models that underly Croston's method (Shenstone and Hyndman, 2005)
There are two variant methods available which apply multiplicative correction factors
to the forecasts that result from the original Croston's method. For the
Syntetos-Boylan approximation (type = "sba"), this factor is \(1 - \alpha / 2\),
and for the Shale-Boylan-Johnston method (type = "sbj"), this factor is
\(1 - \alpha / (2 - \alpha)\), where \(\alpha\) is the smoothing parameter for
the interval SES application.
Croston, J. (1972) "Forecasting and stock control for intermittent demands", Operational Research Quarterly, 23(3), 289-303.
Shenstone, L., and Hyndman, R.J. (2005) "Stochastic models underlying Croston's method for intermittent demand forecasting". Journal of Forecasting, 24, 389-402.
Kourentzes, N. (2014) "On intermittent demand model optimisation and selection". International Journal of Production Economics, 156, 180-190. tools:::Rd_expr_doi("10.1016/j.ijpe.2014.06.007").
library(tsibble)
sim_poisson <- tsibble(
time = yearmonth("2012 Dec") + seq_len(24),
count = rpois(24, lambda = 0.3),
index = time
)
sim_poisson %>%
autoplot(count)
sim_poisson %>%
model(CROSTON(count)) %>%
forecast(h = "2 years") %>%
autoplot(sim_poisson)
Run the code above in your browser using DataLab