nitish wrote:> > I have a dataset that goes like: dataset > t |x > 0 |x1 > 1 |x2 > 2 |0 > 3 |0 > 4 |0 > 5 |0 > 6 |x3 > 7 |0 > 8 |0 > 9 |0 > 10 |x4 > > and so on. I wish to detect the periodicity of occurrences. t is in > seconds and x are arbitrary, whose magnitude i am not interested in. I > just wish to get a best guess for the period of mere occurrences of x. > Tried using spec.pgram(dataset) but the periods I get are too small - of > the order of 1e-5. What am I doing wrong? >$acf Dieter -- View this message in context: http://r.789695.n4.nabble.com/Getting-a-periodogram-for-discrete-data-tp3079658p3079692.html Sent from the R help mailing list archive at Nabble.com.
Thanks Dieter, that solves the next step for me! As per "On Periodicity Detection and Structural Periodic Similarity" y Michail Vlachos, Philip Yu, Vittorio Castelli, the recommended first step is to determine the top 1% of the PSD and then use the range thereof to run autocorrelations, and then loop once to refine the strongest period-candidates. Regards, Nitish -- View this message in context: http://r.789695.n4.nabble.com/Getting-a-periodogram-for-discrete-data-tp3079658p3079693.html Sent from the R help mailing list archive at Nabble.com.