Table 2.
Adjacent Table to Figure 4 .
Group Analyzed | Participants by gender | Participants by Survival | Women by Survival | Men by Survival | ||||
---|---|---|---|---|---|---|---|---|
PC summary | PC1 | PC2 | PC1 | PC2 | PC1 | PC2 | PC1 | PC2 |
Eigenvalue | 9.8 | 7.3 | 9.5 | 7.6 | 10.1 | 9.5 | 7.6 | 4.7 |
Proportion of variance (%) | 30.6 | 22.7 | 29.7 | 23.9 | 31.7 | 29.7 | 23.9 | 14.6 |
Cumulative proportion of variance (%) | 30.6 | 53.3 | 29.7 | 53.6 | 31.7 | 61.3 | 23.9 | 38.5 |
PC1 shows the highest variance of loading on a single vector. PC2 shows the cumulative variance of loading that is orthogonal to PC1 with a center 0. Yet negative variables show the presence of hidden (latent) variables that can be only determined through inference using mathematical modeling or through direct measurement alongside observed variables (those with positive variance on PCA). The greater the eigenvalues than one (1), the greater the predictability power of the variable in determining the hidden or latent variance to the outcome.