Alternate methods of representing single-channel data

Abstract

Clustering of dwell times in data from single-channel recordings, which is in excess of the value predicted from the probability density function (pdf) alone, provides restrictions on modeling schemes. Two methods, (a) the probability density function of the running median for groups of any size of sequential dwell times, and (b) the distribution of cumulative probabilities associated with dwell times separated by any lag, or the second cumulative probability distribution, are proposed as alternative representations of single-channel data; these methods are suitable for the detection of such clusters or modes. Simulation of three models with and without modes is done to test the efficacy of these methods. It is found that they often yield a better estimate of moding parameters than the methods of running mean pdf and autocorrelation.