Macro. Performs a time domain cross-correlation of a series.
CROSSCOR(series1, series2)
series1 |
- |
A series. |
series2 |
- |
A series. |
A table or series.
CONV(S1,REVERSE(S2)/(SERSIZE(S1)+SERSIZE(S2))
W1: gsin(128, 1/128, 4.0)
W2: gsin(128, 1/128, 4.0)
W3: crosscor(W1, W2)
performs a cross-correlation of two sine waves.
W1: grand(128, 1/128)
W2: gsin(128, 1/128, 4.0)
W3: crosscor(W1, W2)
performs the cross-correlation of a sine wave with a random series.
CROSSCOR calculates the cross-correlation directly in the time domain via convolution without normalization. Use XCORR for a time domain version that supports normalization and FXCORR for a frequency domain implementation.
The cross-correlation function is often used to indicate how "similar" one waveform is to another. The cross-correlation of the above sine waves returns a waveform with several distinct peaks, indicating that the two series are very similar at each point in time where the peaks occur.
The cross-correlation of the random series with the sine wave results in a waveform with some peaks, but the amplitude of the waveform is considerably reduced (try OVERPLOT), indicating that the two input series are very dissimilar.
See CORRCOEF to compute the correlation matrix.