Measure Parameter Name Notation

 

A large number of different measures with varying parameters can be generated. To keep track of the measure names unique notation is used for the measure parameters. The full name of each measure is comprised by a 10 character string coding the name of the measure, followed by the measure-specific parameter declaration (one value for each parameter).

Each parameter is identified by a character, as shown in the first column in the table below.

The parameters may bear only a single value or a range of values given in standard matlab syntax, e.g. 1:10 or [1 5:10 20] (see second column in the table).

The parameters may be constrained to specific ranges, as given in the third column of the table. 

 

Parameter Notation
Character Multiple values Range Default

 Description 

a yes integers >0 1 moving average filter order, when smoothing the time series for the detection of turning points
b yes integers >0 0 number of bins when partitioning the time series range, if b=0 then it is determined by the criterion b=sqrt(n)
c yes [0, 1] 0.5 correlation sum, for finding the radius given the correlation sum.
e no integer 2 10 escape factor, for assessing the false nearest neighbors
f no [0.1, 0.9] 0.5 fraction for the test set, e.g. f=0.3 indicates that the test set is 0.3 times the time series length
g no integer 0 0 Theiler's window, to exclude temporally correlated points from the set of neighboring points
h no integers >0 1 lead time for fitting and prediction
k yes integers >0 1 number of nearest neighbors
l no [0, 0.5] multiple cases lower band frequency, for the computation of energy in frequency bands (assuming that the sampling frequency is one)
m yes integers >0 1 embedding dimension or AR model order
o no integers10 100 number of radii for the computation of the correlation sum in the correlation dimension estimation, low resolution (small 'o' facilitates the computation at the cost of accuracy)
p yes integers >0 1 order of the moving average part in ARMA
q no integer 0 0 truncation parameters, q=0 for local average map, q<m for PCR solution of the local linear model parameters, q=m for standard OLS solution of the local linear model parameters
r yes [0, 1] 0.1 radius for the computation of the correlation sum (the time series is standardized in [0,1]).
s yes integers >0 4 upper/lower ratio of scaling window, i.e. s=r2/r1, where r2-r1 is the length of the scaling window
t yes integers >0 1 delay
u no [0, 0.5] multiple cases upper band frequency, for the computation of energy in frequency bands (assuming that the sampling frequency is one). Note that it should hold u>l.
w yes integers >0 1 offset for the local window, the local window has length 2*w+1 and it is used for the detection of turning points