to sum up
Like principal component analysis, we can use factor scores f1 and f2 as two new variables for subsequent modeling (such as clustering, regression, etc.)
Note: The factor analysis model cannot be used for comprehensive evaluation. Although many papers are written in this way, this is a big problem. For example, the type of variable, the method of selecting factors, and the effect of rotation on the final effect are difficult to clarify.
Suggest:
- Linear dimensionality reduction first factor analysis
- Using SPSS software will be much faster
- The data to be multiplied by the factor score must be standardized first (it can be obtained with one click in the descriptive statistics of spss)
- Factor analysis is often better than principal component analysis, because factor loading rotation can be performed, so more solutions can be obtained to carry out the connotation of new factors