Spiel G, Kundi M, Benninger F
EEG EMG Z Elektroenzephalogr Elektromyogr Verwandte Geb. 1986 Mar;17(1):2-6.
Analysis of EEG frequency spectra leads in two fields to problems demanding optimal data reduction methods: especially if topological aspects have to be considered, it is likely that the rather limited memory capacity of laboratory computer systems will be exceeded; furthermore severe methodological problems arise in statistical analysis of combined data sets, including frequency spectra, other physiological, and non-physiological data. Three data reduction methods are discussed: the classification with respect to frequency bands, the search for prominent frequencies, and the computation of quartiles of the frequency spectra. There is some evidence in favour of the method of prominent frequencies. This method seems to preserve much of the differential information of the frequency spectra.
特别是在必须考虑拓扑学方面时，实验室计算机系统相当有限的存储容量很可能会被超出；此外，在对包括频谱、其他生理和非生理数据的组合数据集进行统计分析时会出现严重的方法学问题。本文讨论了三种数据缩减方法：按频段分类、寻找突出频率以及计算频谱四分位数。有一些证据支持突出频率法。该方法似乎保留了频谱的许多差异信息。
