I have a data set which has two patterns of data. One pattern of the data has large variance and other pattern has very less variance. Please suggest an algorithm to find the number of times the pattern has appeared.

If you have a dataset and you want to count the number of times that a given pattern appears in the database, then what about reading each line of your dataset and counting
if the pattern appears in each line?

Your question is a little bit abstract if you don't say what kind of pattern or what kind of data.



Edited 1 time(s). Last edit at 02/06/2014 12:54PM by webmasterphilfv.

I will make it clearer. Following is the format of data.
(1,5)(2,-5)(3,4)(4,2)(5,-3)(6,1.01)(7,1)(8,1.015),(9,1.0095)(10,1.011)..........

This is the sample data points. We can observe that the first five points have a huge variance(pattern 1) and next five points have very less variance(pattern 2). This data is large and will have pattern 1 and 2 alternately. Now, is there an algorithm which will detect how many times the pattern 1 has appeared in the whole data ?

I'm not sure about specific algorithms for that but what you have is called a "time series".

Then for finding patterns, I think that what you want may be called to discover "periodic patterns.

Maybe searching with these keywords could give you some ideas

Yes, it is a time series. But the pattern is not periodic. It does not repeat after a fixed interval of time. For eg. first 5 points have high variance and the next 10 points may have very less variance and next 7 may have high variance and so on...So, to which category of data mining does this type of scenarios goes?

Please let me know if there any tools/ libraries to accomplish this?