Datasets

Here are some datasets used by the group. You will frequently find them in our publications.

• Poland Electricity Load
  • Download
  • Electricity load values of Poland from 1990's
  • Data set size: 1400 values training set and 201 values test set
• Santa Fe Laser time series
  • Link to the source site and Download
  • Recorded from a Far-Infrared-Laser in a chaotic state
  • Data set size: 10000 values
• Darwin sea level pressures
  • Download
  • Monthly values of the Darwin Sea Level Pressure series, 1882-1998
  • Data set size: 1300 values
• Tree-rings
  • Link to the source site and Download
  • Tree-ring widths in dimensionless units, in Argentina, year 441-1974
  • Data set size: 1534 values
• Atmospheric CO2 from continuous air samples
  • Download
  • Weekly averages atmospheric CO2 concentration derived from continuous air samples, Hawaii
  • Data set size: 2225 values
• Quebec births dataset
  • Download
  • Number of daily birth in Quebec, from Jan 1, 1977 to Dec 31, 1990
  • Data set size: 5113 values
• Daily minimum temperatures
  • Download
  • Daily minimum temperatures in Melbourne during 1981-1990
  • Data set size: 3650 values
• Electricity Demand
  • Download
  • 15 minutes averaged values of power demand in the full year 1997
  • Data set size: 37440 values
• Subtidal coastal level of fluctuations
  • Download
  • Analysis of Subtidal Coastal Sea Level fluctuations Using Wavelets, 1997
  • Data set size: 8746 values
• Sunspot Number
  • Link to the source site and Download
  • Monthly averages of sunspot numbers from 1749 through the present
  • Data set size: 3078 values
• CATS benchmark
  • Link to the source site and Download
  • This artificial Time Series with 5,000 data is given. Within those 100 values are missing. These missing values are divided in 5 blocks
  • Data set size: 5000 values
• Anthrokids (without missing values)
  • Download
  • Anthropometric study of US children carried out in 1977. The objective is to predict their weight. This a modification of the original data set from http://ovrt.nist.gov/projects/anthrokids without missing values. The data are normalized.
  • 1019 samples and 55 variables, normalized
• Finance
  • Download
  • Information about 200 French industries during a period of 5 years. The variables are related to balance sheet, income statement and market data, and the output is the return of assets (ROA). The data are normalized.
  • 650 samples and 35 variables, normalized
• ESTSP 2007 Competition
  • Download
  • Time series proposed for the European Symposium on Time Series Prediction 2007
  • 875 measurements
• Finance 2
  • Download
  • Time series about internal financial dataset concerning fund returns
  • 120 time series with 121 measurements
• ESTSP 2008 Competition: Data 1
  • Download
  • First time series proposed for the European Symposium on Time Series Prediction 2008
  • 354 measurements
• ESTSP 2008 Competition: Data 2
  • Download
  • Second time series proposed for the European Symposium on Time Series Prediction 2008
  • 1300 measurements
• ESTSP 2008 Competition: Data 3
  • Download
  • Third time series proposed for the European Symposium on Time Series Prediction 2008
  • 31614 measurements
• Financial Ratios, Bankruptcy Prediction
  • Download
  • Financial indicators from several anonymised companies from 2002 for a bankruptcy prediction task (1 = healthy, 2 = bankrupt).
  • 500 samples with 40 variables