Verschuure J, van den Wijngaart W S, Brocaar M P, Nagels M M
Audiology. 1985;24(1):2-14. doi: 10.3109/00206098509070092.
The analysis of large numbers of audiograms raises the question if and how we can reduce the amount of data without discarding essential information. The present paper compares two ways of data reduction: principal-component analysis and curve fitting. The methods are tested on the audiograms of a large family suffering from a dominant hereditary, progressive hearing loss, beginning in the high frequencies. It is shown that principal-component analysis rejects information on the shape of the audiogram, as do all methods generally referred to as factor analysis. The information concerned is essential for our understanding of the patient's ability to discriminate speech. Curve-fitting procedures are shown to be effective in data reduction.
我们是否能够以及如何在不丢弃重要信息的情况下减少数据量。本文比较了两种数据简化方法：主成分分析和曲线拟合。这些方法在一个患有显性遗传性进行性听力损失（始于高频）的大家族的听力图上进行了测试。结果表明，主成分分析会舍弃听力图形状的信息，所有通常称为因子分析的方法也是如此。所涉及的信息对于我们理解患者辨别语音的能力至关重要。结果表明，曲线拟合程序在数据简化方面是有效的。
