主成分分析是因子分析的一种特例，因子分析往往更容易建立模型与数据之间的直观的具有明显意义的关系。

尤其在因变量比较多的情况下，因子分析效果更加直观，更好。

能说说你具体要研究的是什么问题嘛？

• Spectroscopy and PCA
• An introductory explanation of PCA from StatSoft
• A Tutorial on Principal Component Analysis (PDF)
• A tutorial on PCA by Lindsay I. Smith (PDF)
• A layman's explanation from Umetrics
• Principal Component Analysis using Hebbian learning tutorial
• Presentation of Principal Component Analysis used in Biomedical Engineering
• PCA in functional neuroimaging, free software

Principal components analysis (PCA) is a technique used to reduce multidimensional data sets to lower dimensions for analysis. Depending on the field of application, it is also named the discrete Karhunen-Loève transform (or KLT, named after Kari Karhunen and Michel Loève), the Hotelling transform (in honor of Harold Hotelling), or proper orthogonal decomposition (POD).

l The PC transformation should be applied to data that have approximately the same scale in each variable (A.A.Afifi: Computer-Aided Multivariate Analysis.CRC.1997)

l As with all statistical analysis, interpreting data from a random sample of a well-defined population is simpler than interesting data from a sample that is taken in a haphazard fashion.

l The data used for PC should follow a multivariate normal distribution.

l If PC is used to check the redundancy in the data set, it is very important to have observation measured accurately.

l Retain sufficient components to account for a specified percentage of the total variance say 80%(Wolfgang Hardle: Applied Multivariate Statistical Analysis, Wiley 2007)

l Retain components whose eignvalue is larger than average eigenvalue (Tanu Kollo: Advanced Multivariate Statistics with Matrices, Springer)

l Use the Scree Graph to look for a natural break between the larger eigenvalue and small eignvalue.

l Use the scandalized data(Subhash Sharma: Applied Multivariate Techniques, Wiley 1996).

l 当分析中所选择的经济变量具有不同的量纲，变量水平差异很大，应该选择基于相关系数矩阵的主成分分析。

[此贴子已经被作者于2007-9-3 6:07:56编辑过]