
	Keywords: Spike detection; Spike sorting; Spike train analysis; Extra-cellular recording; Tetrode; Electrophysiology; Multi-unit recording; Single-unit recording; Cluster analysis; Superposition resolution; MATLAB; Accuracy quantification; Object Oriented Programming; Data simulation; Neuroshare Native

StationaryTest Tool (Under Development)

Summary

Stationarity is the fundamental assumption behind all of the point process analyses found in the Spike Train Analysis toolbox. When a stochastic process is "stationary" the stochastic description of the process is independent of time. If a process is stationary, the short term averages computed from non-overlapping blocks should be a stationary Gaussian process by the central limit theorem. The StationarityTest performs several statistical analyses that test this condition.

Screen Shot

Details

The StationarityTest computes short term averages and then uses several tests to ask whether or not the samples are consistent with a Gaussian model.

The tests used include:
"lillietest" - Use the "Lillifors" test; more information in "help lillitest"
"jbtest" - Should not be used for small samples; more information at "help jbtest"
"vartest" -- Based on stationarity criterion outlined in (D.H. Johnson, 1996 and D.H. Johnson, et al., 1986). Tests whether or not the average interval computed for the entire data sets deviates significantly from the short term averages.

For all these hypothesis tests, the null hypothesis is that the data is normally distributed with unknown mean and variance. All hypothesis tests return a value of "0" if you cannot reject the null hypothesis, "1" if you can reject it at a significance level of alpha * 100%.

R.C. Electronics Inc. 6464 Hollister Avenue Santa Barbara, CA 93117
Phone: (805) 685-7770 Email: info@neuromax.org

Last Updated: 25-Mar-2008

USERID:
PASSWORD:
	Forget Password?