Stat 147 Class Activity on PCA

TRUE or FALSE: Doing PCA on the correlation matrix of the random vector is equivalent to doing PCA on the standardized version of the random vector.

• TRUE

• FALSE

TRUE or FALSE: In Single Value Decomposition, the principal components are derived directly from the correlation matrix of the dataset.

• TRUE

• FALSE

TRUE or FALSE: If we conduct PCA on a random vector that follows multivariate normal, then the principal components will be independent.

• TRUE

• FALSE

• TRUE

• FALSE

The sum of the variances of all the principal components derived from the covariance matrix is equal to ____________.

• 1

• p

If we are conducting PCA based on the correlation matrix, then according to Kaiser's Rule (with correction from Joliffe), we retain the PCs wherein ____________.

• The eigenvalues are greater than or equal to 0.7.

• The eigenvectors are normalized.

• The eigenvalues are lower than the average of the eigenvalues.

• There is an elbow in the Scree Plot.

• When there is little correlation between the original variables

If we are doing a PCA on the correlation matrix, when do we encounter a case wherein at least one of the principal components have zero variance-explained from the original dataset?

• When the correlation matrix is equal to the identity matrix

• When the correlation matrix is not of full rank

• When the correlation of the original variables are low

• When the original variables are normalized