When considering time as a variable which undergoes cycles across a data-series (ex. Month of year across a multi-year time-series, or time of day in a multi-day study), it is often useful to express time on a circular scale. Typically, I see this accomplished with a sine or cosine transformation of the data. While this approach does indeed circularize the data, when a vector of evenly spaced values is transformed, values in the resultant vector are not evenly spaced. How would one best circularize a time vector, but retain the interval spacing between points?
Hi, as you transform the orginal data via a cosine for example, if you want at the end a uniform spacing between the outputs you need to generate acos of a grid before applying the cosine transform.
Assuming you want to pick periodic( say M) elements from a given data the Comb function in DSP is a likely candidate. Please see attached pdf .
Cheers
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication