Hi,

Assume your dataset as X. at first, assign a label to subset of X according to event. label can be represent as a matrix Y.

different  classifiers can be used for your problem. the classifier is like Y=f(X) . 

some classifier in Matlab:

Neural network

Gaussian Naive bay

Support vector machine

After training the classifier (tune the parameter of f) you can use new dataset to test the performance of the classifier.

it should be consider the standard classifier does not consider the time sequence in samples. Indeed they just try to classify each snapshot of test data set.

Thanks Mahdieh for your reply

I wanted to know how to classify time series data. Is there any algorithm which does it or just the trend checking on a graph is sufficient.

you can extract time series features (such as entropy, power distribustion). Then use these features to train the classifiers.

It is common to use rolling window to explore the time feature of test dataset.

Thanks Mahdieh for your prompt reply. I would try to implement as you said. I may disturb you again in future and hope you would not mind.