Skip to main content

Table 2 Sparsity detection, example 2.

From: The projection score - an evaluation criterion for variable subset selection in PCA visualization

Measure First component Second component
Projection score 160 (156-160) 152 (133-156)
Gap statistic 122.5 (106-133) 78 (60-100)
SPC 260.5 (229-499) 311.5 (280-494)
SSVD 162.5 (161-339) 164 (162-313)
  1. The median and range (across 10 instances) of the number of non-zero elements included in the optimal variable subset found by different methods. Each underlying component contains 160 non-zero elements. SPC - sparse PCA [22]. SSVD - sparse SVD [27]. The gap statistic is defined as in [25, 26].