Almost all datasets in the UCR archive contain only time series sequences that are segmented and scaled to the same length, which is unrealistic. Are there any time series datasets that are not segmented but labeled?
1
migrated from datascience.stackexchange.com Dec 19 '18 at 14:32
This question came from our site for Data science professionals, Machine Learning specialists, and those interested in learning more about the field.
1
The new UCR datasets have a few such examples, see https://arxiv.org/abs/1810.07758
Also, there are three or four such datasets in https://www.cs.ucr.edu/~eamonn/SDM_RealisticTSClassifcation_cameraReady.pdf
eamonn
