# 加拿大论文代写：因子分析

PCA has close relation with the analysis of factor. Analysis of factor, in the typical sense, incorporates assumptions specific to domain regarding the underlying structure, while solving eigenvectors over a matrix that is slightly different. In addition, the results obtained from PCA have key dependence upon the variables scales, while there is a development of PCA in form of invariant scale (Abdi & Williams, 2010). There is limitation of applicable PCA by specific assumption while drafting out its derivation.
The first presumption is in terms of linearity that data set will be linear combinations of each and every variable involved. The second presumption is regarding the significance of co- variance and mean. There does not seem to any guarantee in directing maximum variance that consists of good attributes for discrimination (Bowen & Guo, 2011). The third presumption is about large variances having significant dynamics.
This states that components having large variance will be corresponding with the crucial dynamics, while the lower ones will be corresponding with noise. In statistical terms, there is use of varimax rotation for simplifying the expression of a specific sub- space with respect to some major items. The actual system of coordinate will not change. No change will take place as it is the orthogonal base, rotating for ensuring alignment between different coordinates (Bro & Smilde, 2014). The discovery of sub- space with PCA can be expressed as a dense base with a number of non- zero weight, creating difficult interpretation. Varimax is needed as rotation as it helps in maximizing the total variances across the squared loadings. Squared loading can be identified as squared correlations between factors and variables. The preservation of orthogonality requires that the rotation will be leaving the invariant of sub- space.