J E Beasley

OR-Library is a collection of test data sets for a variety of OR problems.

A full list of the test data sets available in OR-Library can be found here.

Time series forecasting

There are currently 4 data files.

The first two data files (time1 and time2) contain the problems 
solved in S.Makridakis et al "The accuracy of extrapolative (time 
series) methods: results of a forecasting competition" Journal of 
Forecasting 1 (1982) 111-153.

File time1 has 1001 data series and file time2 has 111 data series.

The third data file (time3) contains the problems solved in R.Fildes
"The evaluation of extrapolative forecasting methods" International
Journal of Forecasting 8 (1992) 81-98.

File time3 has 263 data series.
These data files were contributed by Professor Robert Fildes of
Lancaster University.
The fourth data file (time4) has the data used in 
"Modelling multinational telecommunications demand with limited data"
by T. Islam, D. Fiebig and N. Meade (email n.meade@ic.ac.uk). 
The paper has been submitted to International Journal of Forecasting.
Four data sets are used, each concerning the growth of demand for a 
telecom service.
The format of time1 and time2 is:
number of series (S)
for each series s (s=1,...,S) in turn:
number for the series
number of data points (n)
number of forecasts ahead (6 for annual, 8 for quarterly,
                           18 for monthly)
number for the series (equal to the previous series number)
data points in the series (n numbers)
the forecast data points (6, 8 or 18 numbers as above)
seasonal factors (12 numbers with the first number being 1 if non-seasonal)

The format of time3 is:
number of series (S)
for each series s (s=1,...,S) in turn:
name for the series
data points in the series (71 points for each series)
The format of time4 is given in the data file.

The largest file is time1 of size 1300Kb (approximately)
The entire set of files is of size 1700Kb (approximately).

Click here to access these files

The M series competition data can be obtained here