| Title: | Data from the 2010 Tourism Forecasting Competition |
|---|---|
| Description: | The 1311 time series from the tourism forecasting competition conducted in 2010 and described in Athanasopoulos et al. (2011) <DOI:10.1016/j.ijforecast.2010.04.009>. |
| Authors: | Peter Ellis [aut, cre] |
| Maintainer: | Peter Ellis <[email protected]> |
| License: | GPL-3 |
| Version: | 1.0.1 |
| Built: | 2024-10-31 03:04:38 UTC |
| Source: | https://github.com/ellisp/tcomp-r-package |
Applies four modelling strategies (ARIMA, ETS, Theta and naive or seasonally naive) to a dataset
with class Mdata, returns accuracy statistics and (optionally) a summary graphic
forecast_comp(the_series, tests = list(the_series$h), plot = FALSE, ...)forecast_comp(the_series, tests = list(the_series$h), plot = FALSE, ...)
the_series |
a list of class |
tests |
a list of the forecast horizons over which to return the MAPE and MASE, passed to |
plot |
whether or not to draw basic plot of the four forecast model |
... |
other parameters to pass to |
This is just a convenience function for fitting four different standard time series forecasts to an object from an M competition or the tourism competition. Mainly of interest to analysis wishing to re-produce published results or create a benchmark against which other methods can be compared.
A data frame of eight rows and length(tests) + 1 columns with first column as method and each other column containing the
Mean Absolute Percentage Error (MAPE, first four rows) and Mean Absolute Scaled Error (MASE, final four rows) at the horizon
indicated by the column name
forecast_comp(tourism$Y18, test = list(1, 2, 3, 4, 1:2, 1:4), plot = TRUE) forecast_comp(tourism$Q4, test = list(1, 2, 3, 4, 5, 6, 7, 8, 1:4, 1:8), plot = FALSE)forecast_comp(tourism$Y18, test = list(1, 2, 3, 4, 1:2, 1:4), plot = TRUE) forecast_comp(tourism$Q4, test = list(1, 2, 3, 4, 5, 6, 7, 8, 1:4, 1:8), plot = FALSE)
The Tcomp package provides data from the tourism forecasting competition described in George Athanasopolous, Rob J. Hyndman, Haiyan Song, Doris C. Wu (2011) “The tourism forecasting competition”, International Journal of Forecasting 27 (2011) 822-844.
A list of class Mdata with 1,311 time series in it, divided into
training (x) and test (xx) sets. The data are subsettable with subset method from Mcomp R package by monthly, quarterly and yearly series.
A convenient wrapper function for providing the mean absolute scaled error (MASE) of four common forecasting methodologies for a time series of class Mdata.
Reproduction of selected results from the tourism forecasting competition described in Athanasopoulos et al. 2011 (http://robjhyndman.com/papers/forecompijf.pdf)
Tcomp_reproductionTcomp_reproduction
A list of three elements named monthly, quarterly and yearly.
These correspond to tables 4, 5 and 6 in the Athanasopoulos et al 2011 article.
Note that only Mean Absolute Percentage Error of the naive forecasts matches exactly that published. All Mean Absolute Scaled Error results are slightly higher than those published due to an unknown difference in MASE method. All results for ARIMA, ETS and Theta method forecasts differ due to changes in the forecasting methods since 2011. See Vignette for details, including the code required to re-create the 'Tcomp_reproduction' object.
http://robjhyndman.com/papers/the-tourism-forecasting-competition
Tcomp_reproductionTcomp_reproduction
The data from the tourism forecasting competition described in George Athanasopolous, Rob J. Hyndman, Haiyan Song, Doris C. Wu (2011) “The tourism forecasting competition”, International Journal of Forecasting 27 (2011) 822-844.
tourismtourism
A list of 1311 series, of class Mcomp. Each series within tourism is of
class Mdata with the following structure:
Name of the series
Series number and period. For example "Y1" denotes first yearly series, "Q20" denotes 20th quarterly series and so on.
The number of observations in the time series
The number of required forecasts
Interval of the time series. Possible values are "YEARLY", "QUARTERLY" & "MONTHLY"
The type of series. For data in tourism, this is always "TOURISM".
"No description available". Kept for consistency with the M3 and M1 data.
A time series of length n (the historical data)
A time series of length h (the future data)
http://robjhyndman.com/papers/the-tourism-forecasting-competition
plot(tourism$Y1)plot(tourism$Y1)