PCA+Assumptions

PCA Assumptions (Principal Components Analysis) (GWU EMSE-271)
Index | Topics (Logical Lectures) | Lectures | Problems | Readings | Nomenclature | Concepts

Not finding anything quickly in Wikipedia.

Lattin does not have a nice, bold assumptions box like at the start of regession analysis.

> - statistical importance of mean and covariance > - large variances have important dynamics - Wikipedia > - Not independent" – Lattin, pg 122
 * 2011/1/22: Found this in my notes: " - Linearity

Lattin has a little on model validity (but validity on pae 117 is talking about whether we can generalize.

Lattin's learning points help a little:
 * Not always necessary to standarize, but it can be highly sensitive to scale differences.
 * Highly susceptible to outliers and influential observations. (That's why its good to use holdout options (like (below)) to validate the solution.

Not hypotheis test found in lectures (assume no assumptions to check)
 * ** Assess Validity ** ||  || Don't remember being covered in Lecture for PCA. ||
 * - Jackknife Validation ||  ||   ||
 * - Bootstrap Validation ||


 * Sources:**
 * EMSE 271, Fall 2009