Hi all,
Is there any multivariate time series classification problem in which some variables are categorical? The time series are assumed to have numerical observations for most of the approaches in this domain and I am interested in the case where some observations are categorical.
For example, network flow data consists of packages transferred between IP pairs. Each flow can be labeled by its application such as bittorrent, Skype and etc. Each flow is a series of packages for which the information about the size, direction and payload is known. Direction is either upstream or downstream in this particular example. Although it can be represented as a binary variable, the natüre of the variable is categorical.
Please, let me know if you have such datasets. Thanks in advance.
Hi,
Here there are some EEG based brain computer interface data sets links:
1- http://eee.ktu.edu.tr/temp/Datasets/contentlister.asp
2-http://www.bbci.de/competition/iii/
3- http://www.bbci.de/competition/ii/
Hi Önder,
Thanks for your answer but these datasets do not have categorical variables. I have plenty of datasets with numerical observations mostly from motion capture studies.
Best regards,
Couldn't you simulate such datasets? It seems like you'd want to know exactly what the data generating process entails so that you can see how well the model actually estimates the results.
That is something doable. For example, I can generate synthetic series with hidden markov models. Different parameters can imply different classes. However I am interested in real life problems.
If anybody requires the type of dataset mentioned in this post, you can download it through my website. Network flow dataset in this repository stores the information about the packages sent by applications such as skype, bittorrent and so on. The aim is to predict the application type based on the set of packages in the flow.
The time series dataset stores the information about the direction of the package at a given time, time between packages, size of the package and size of the payload. This information is collected at Bogazici University by Subakan et. al. (Y. C. Subakan, B. Kurt, A. T. Cemgil, and B. Sankur. Probabilistic sequence clustering with spectral learning. Digital Signal Processing, 29(0):1 – 19, 2014.).
Here is the link to the study introducing this dataset: http://www.busim.ee.boun.edu.tr/~sankur/SankurFolder/DSP_SpectralLearning_Subakan.pdf
http://www.mustafabaydogan.com/blog/15-learned-pattern-similarity/75-new-multivariate-time-series-classification-datasets-are-added-to-files-section.html
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series