Time series has similar patterns. A pattern consists of several segments. Each segment could be represented as a simple parametric regression function. Thewhole pattern is the sequence of concantenated regression functions.The following information is given: One-dimensional time series and a set of the regression functions.Each function defines:
- type of regression function (linear, exponential, gaussian, etc.);
- parameters of the function;
- minimal and maximal number of samples in the segment;
- maximal value of the variance of the residuals.
One must compute starts of the segments. See the demo.