I want to classify some data. Basically the data is time series in nature. The target variable is categorical. I know there are so many algorithms for predicting the time series model. However, I have very little information on classifying the time series model. Can anyone tell me how to classify the time series model based on categorical data ? What kind of approach will solve the problem ? I have used HMM for classification of part of speech tagging, entities. But it is not time dependent. I found that there are good algorithms available using kNN and Dynamic Time Warping. Is this the conventional model used for time-series classification ? Any other model available ? Your thoughts much appreciated.
Well.. your features should represent always the same thing. Say you have 1 second time-series with a sampling frequency of 100 Hz, so 100 samples per second. Then you have some number of such time-series. In each of the time-series, does x(t) represent exactly the same thing? If not, I would first transform the time-series into the frequency domain. That can be done by taking the Fourier transform (FT). Does the FT reveal any interesting shapes, e.g. peaks at certain frequencies? Limit your analysis to those. The classifier can then be any I would say, picking the correct data is more important.