Splining through glitches

[ top | up ]

Splining Through Glitches

Syntax

[ntimes,newvals,newlabels] = hrspline(times,vals,labs,goodlabel);

Arguments

`times`	A column vector giving the time at which each of the `vals` was recorded.
`vals`	A column vector of the recorded data.
`labels`	A column vector indicating which of the `vals` are valid.
`goodlabel`	A number indicating what label corresponds to valid. Default: 1.

Returned Values

`ntimes`	Just a copy of `times`
`nvals`	A copy of `vals`, but with the invalid points replaced by a linear spline between the adjacent valid points.
`newlabels`	A copy of `labels`, with 1s whereever `labels` indicated a valid point, and 0s elsewhere.

Description

It is often necessary to remove invalid data from a data set. In some cases, e.g., power spectrum analysis, one must at the same time maintain the original timing of the valid data points. Splining is one technique for doing this; it replaces each invalid data point with a value derived from the nearest valid data points on each side by computing the value that a linear spline connecting those valid data points would have at the time of the invalid point under consideration. Invalid data points at the extreme ends of the time series are replaced with the value of the nearest valid point.

References

Albrecht's paper in Computers in Cardiology