Kim Kuen Tang wrote:
Hi Eric,
i just want to point out that Boost.Time_series is not compatible with the current boost release 1.36. The compilation of Boost.time_series together with boost 1.36 was not successful due to some errors.
Thanks for the heads up. I'll look into it.
You have written in your description that Boost.Time_series does not yet contain all the algorithms one might want in order to perform full time series analysis. Does it mean that the user should choose Boost.time_series as a starting point to implement the algorithm to perform full time series analysis?
That's correct. Although what is there may be sufficient for many uses, Boost.Time_series is mostly about getting the concepts correct and the framework in place in the hopes that the "community", or at least one or two other interested people, can help fill in the missing pieces.
Let's say i want to implement an algorithm to calculate the spectral density on a time series. Why is Boost.time_series a good choice for me?
How is your time series represented? Sparse or dense or piecewise constant or ...? Would you like to implement the algorithm once and have it be maximally efficient for all of these different series representations? Then Boost.Time_series is a good choice.
Since the library Boost.accumulators already contains many useful functions like count, covariance,.. I would choose it as my starting point.
If you're only dealing with a dense series, then this is a reasonable approach.
Thanks for your answer,
My pleasure. -- Eric Niebler BoostPro Computing www.boostpro.com