Counts are variable in M, where M are N times.
Updated: I plotted data and it appeared as Poisson process. But as counts are varying in M. It might be non-Poisson process. But not sure about the how many and what test on non-Poisson data. A kind support needed on such.
Dear Harpreet
For this purpose you need methods that are suitable for streaming data.
Updateable kind of statistical algorithms are needed for use.
Also there are so many methods available in literature.
Cheers
Thanks Dr Indrajit.
May I know any example on "updateable kind" of statistical algorithms?
skew is used to test sideness of distributions.
is your distribution stationary, u need to check that first (i.e. are the parameters of the distribution changing over time or not). if they not changing (u can compare the parameters computed over different time windows, and get their variance etc), u can calculate skew to check if it is right sided.
if it is non stationary, u need to check if it converges to a value (again using the time window computations from above) or if it is fluctuating, or are there trends.
for updates, there are several 'learning' algorithms such as LMS filter, Kalman, etc, for point estimation, as well as other Bayesian techniques as well.
- Badges
- Science topic