webmaster: Sven F. Crone
Centre for Forecasting
Lancaster LA1 4YF
eMail sven dot crone (at)
neural-forecasting dot com
& Branch Summary
On these pages you will find
information & links to data sources necessary to evaluate
forecasting performance of neural networks objectively against other
researchers and practitioners using NN or other methods. Wish to add something
to this portal ? Link exchange only in NN & forecasting sites! Send
info, links etc. to
Data Sources for Neural Forecasting
A variety of sources for data exists for forecasting and
prediction in various domains and application areas - NN are hardly specific to
Below, please find a
selection of the most prominent ones.
Forecasting Competition Data
Most notably, the data from the
three Makridakis & Wheelwright competitions on (univariate) time series
One focus of research on business forecasting has been empirical
validation. Over the past two decades, Spiros Makridakis and Michele
Hibon have led groups of researchers from all over the world in
conducting comparisons of various forecasting methods. Some of the
results of these studies have been surprising. For example, simple
methods seem to do at least as well as more complex methods. And,
combining forecasts helps, particularly for longer horizons. If your
favourite method was not included among those studied, you disagree that
the metrics used were right, or you simply don't like the conclusions of
the studies, you can conduct your own replication or extension. Data
from the M-Competitions has been used by hundreds of researchers.
Recently, specific competitions for
neural networks & other soft computing methods were initiated, most
recently the CATS competition. Please find additional information on the
Please follow the links on the left ...
Time Series Data
Time Series Data Library contains some 500 time series organized
by subject as well as pointers to other sources of time series available
on the Internet. This is by far the best source ...
Micro- and Macroeconomic Data
Three websites guide practitioners, researchers, and
students to data. The data can be used for econometric methods or
analogous situations, or to adjust data such as to account for inflation
access to regional, state, and local economic data. It is a
comprehensive first stop for searching public and private data
sources on the Web.
Statistical Resources on the Web
(University of Michigan) has a detailed A-Z index and provides
thousands of links to sites in business, economics, transportation,
weather, and 18 other areas.
Social Science Data on the Net
(University of California, San Diego) allows you to search the
following sites. Much of the data can be downloaded.
- Census Bureau
- Data FERRET (Current Population
Survey and related sources)
- FedStats (Statistics from 70 U.S.
- ICPSR (Inter-University
Consortium for Political and Social Research)
- Stat-USA (Current and historical
financial and economic data
The UCI Machine Learning Data