Welcome to the Monash, UEA & UCR
Time Series Extrinsic Regression Repository

monash uea ucr

This website aims to support research into Time Series Extrinsic Regression (TSER), a regression task of which the aim is to learn the relationship between a time series and a continuous scalar variable a task closely related to time series classification (TSC), which aims to learn the relationship between a time series and a categorical class label. We recommend you to read the paper for a detailed discussion of the datasets and their sources. If you use the results or code, please cite the paper "Chang Wei Tan, Christoph Bergmeir, Francois Petitjean, Geoffrey I. Webb, Time Series Extrinsic Regression: Predicting numeric values from time series data".

            @article{
              Tan2020TSER,
              title={Time Series Extrinsic Regression}, 
              author={Tan, Chang Wei and Bergmeir, Christoph and Petitjean, Francois and Webb, Geoffrey I},
              journal={Data Mining and Knowledge Discovery},
              pages={1--29},
              year={2021},
              publisher={Springer},
              doi={https://doi.org/10.1007/s10618-021-00745-9}
            }
          

If you just use the website, please reference the website as:

Chang Wei Tan, Christoph Bergmeir, Francois Petitjean, Daniel Schmidt, Geoffrey I. Webb, Anthony Bagnall, & Eamonn Keogh (2020). The Monash, UEA & UCR Time Series Extrinsic Regression Archive. URL http://tseregression.org/.
March 2021 Updated the LiveFuelMoistureContent dataset.
March 2021 The paper "Time Series Extrinsic Regression" is now published in Data Mining and Knowledge Discovery.
December 2020 The paper "Time Series Extrinsic Regression" is now in Press.

Datasets

The following table shows a list of time series extrinsic regression datasets. You can download the entire spreadsheet displayed below here and the whole dataset here (about 600 MB). The datasets are available in sktime .ts format. An example of loading the data can be found in our github repository.

The results page shows some baseline on these data using typical regressors.

Loading, please wait
ID
Type
Dataset
Train Size
Test Size
Length
Dimension
Missing Values
Donor/Source
1Energy MonitoringAppliancesEnergy964214424NoLuis Candanedo (UCI Repository)
2Energy MonitoringHouseholdPowerConsumption174669414405YesGeorges Hebrail & Alice Berard (UCI Repository)
3Energy MonitoringHouseholdPowerConsumption274669414405YesGeorges Hebrail & Alice Berard (UCI Repository)
4Environment MonitoringBenzeneConcentration343354452408YesSaverio De Vito (UCI Repository)
5Environment MonitoringBeijingPM25Quality124325100249YesSong Xi Chen (UCI Repository)
6Environment MonitoringBeijingPM10Quality124325100249YesSong Xi Chen (UCI Repository)
7Environment MonitoringLiveFuelMoistureContent349315103657NoLiuJun Zhu
8Environment MonitoringFloodModeling14712022661NoJihane Elyahyioui
9Environment MonitoringFloodModeling23891672661NoJihane Elyahyioui
10Environment MonitoringFloodModeling34291842661NoJihane Elyahyioui
11Environment MonitoringAustraliaRainfall11218648081243NoBureau of Meteorology Australia
12Health MonitoringPPGDalia4321521482256-512*4NoAttila Reiss, Ina Indlekofer & Philip Schmidt (UCI Repository)
13Health MonitoringIEEEPPG1768132810005NoZhilin Zhang (IEEE Signal Processing Cup 2015)
14Health MonitoringBIDMCRR5471239940002NoPeter Charlton & Marco Pimentel (PhysioNet)
15Health MonitoringBIDMCHR5550239940002NoPeter Charlton & Marco Pimentel (PhysioNet)
16Health MonitoringBIDMCSpO25550239940002NoPeter Charlton & Marco Pimentel (PhysioNet)
17Sentiment AnalysisNewsHeadlineSentiment58213249511443NoNuno Moniz & Luís Torgo (UCI Repository)
18Sentiment AnalysisNewsTitleSentiment58213249511443NoNuno Moniz & Luís Torgo (UCI Repository)
19ForecastingCovid3Month14061841Nocovid19.who.int

* These datasets have equal length series but differs between dimensions.

Results

cd
Critical difference diagram for 13 algorithms. Solid bar means that there is no significant difference in rank between methods (see Demsar 2006). Tests are performed with the two-tailed Nemenyi test.

The following is the raw result for each of the dataset and regressor. The same results can be obtained from here

Loading, please wait
Dataset Name
FPCR
FPCR-Bspline
SVR Optimised
RandomForest
XGBoost
1-NN-ED
5-NN-ED
1-NN-DTWD
5-NN-DTWD
Rocket
FCN
ResNet
InceptionNetwork
AppliancesEnergy5.4050525.4050523.4545743.45511983.4890245.2319534.2274386.0365474.0198732.29903122.8656843.0650474.43533
AustraliaRainfall8.4363358.4363368.6508568.3895418.49298630.25413910.23284112.00198111.950738.1241373338.4258748.1791738.841251
BeijingPM10Quality99.72594699.732125110.57422694.07234493.138127139.22979115.669411139.134908115.502744120.057764694.34872995.48937496.749997
BeijingPM25Quality69.37921769.36989271.43707663.30142859.49586588.19354574.15638288.25608272.71768962.76965559.72684764.46274662.227924
BenzeneConcentration11.08839611.0949744.7909010.8555590.63772566.5356855.844984.9835784.8684653.3606144.9882954.06126081.584852
BIDMC32HR13.98055813.98059713.3929715.01646813.96379914.83650614.75608815.2910115.12700813.944382813.13066510.741429.424679
BIDMC32RR3.3647773.3647043.173664.3503144.3678284.3873454.1346853.5291113.4322474.09290063.5777753.9212143.018405
BIDMC32SpO24.9535194.9535174.7968554.5702624.4508055.5302025.4078755.2150275.1239645.2217375.9683375.9878325.57612
Covid3Month0.0449120.0449120.065840.04240.0446820.053060.0418150.0527350.0429430.04387820.074340.0953380.053769
FloodModeling10.0188530.0188530.0463040.0158910.01597120.014820.0161930.0116890.0098010.0023560.0067090.0088680.01743
FloodModeling20.0190790.0190790.0758040.0140950.0181990.0185520.0185860.0163560.0162380.0058810.0067190.0139390.00729
FloodModeling30.0214580.0214580.0350320.0204290.02070380.0199470.0207650.013750.0133370.0040640.0078730.015580.00821
HouseholdPowerConsumption1147.548998147.5492152.391358248.858964231.089829473.932736432.594707427.04311297.221675132.798779162.244492193.207281153.716402
HouseholdPowerConsumption246.92518546.92978355.9808346.93213944.372932671.47936964.27295658.80263451.49496932.60710446.82925639.08012139.409826
IEEEPPG31.38121431.38121237.25414632.1090731.48790133.20886227.11121337.14039333.57278636.515489234.32572833.15098523.903929
LFMC37.68385737.68807439.73352732.162625232.44188647.83679838.53552639.97170735.18530129.409753833.2572230.351656428.796294
NewsHeadlineSentiment0.1422730.1422720.1429170.1475820.1424860.2028210.1566360.1979370.1558390.1422440.1480650.1500240.150014
NewsTitleSentiment0.1381260.1381260.1388810.1431030.1383360.1933180.150950.1872570.1505640.1380590.1380820.1382950.158558
PPGDalia20.67448820.67448619.00521617.53062816.5827321.87656718.28227726.02457620.76838914.05054413.03880511.3821659.923701

Publications

  1. C. Tan, C. Bergmeir, F. Petitjean, and G. Webb, "Time Series Extrinsic Regression: Predicting numeric values from time series data" in Data Mining and Knowledge Discovery 2021 doi
  2. C. Tan, C. Bergmeir, F. Petitjean, and G. Webb, "Monash University, UEA, UCR Time Series Extrinsic Regression Archive"

About Us

We are a group of time series researchers from Monash University, University of East Anglia and University of California Riverside:

changwei
Chang Wei Tan
tony
Anthony Bagnall
christoph
Christoph Bergmeir
daniel
Daniel Schmidt
eamonn
Eamonn Keogh
francois
François Petitjean
geoff
Geoff Webb