Skip to main content

Table 5 Average misclassification rates and number of variables selected for the depression study

From: Sparse sliced inverse regression for high dimensional data analysis

Method

Mean test error

Mean

Mean

Selected metabolites (SE)

Sensitivity (%)

Specificity (%)

\({\widehat{{\varvec{\beta }}}}\)

CISESIR

0.1955 (0.0150)

84.62 (2.63)

72.27 (3.71)

40.540 (2.634)

CISELDA

0.1659 (0.0095)

80.37 (2.32)

83.20 (2.32)

26.280 (1.828)

MGSDA

0.1639 (0.0093)

84.62 (1.36)

81.47 (2.02)

16.129 (0.772)

PLDA

0.2484 (0.0093)

89.62 (0.97)

59.73 (2.36)

78.200 (2.797)

MSDA

0.1458 (0.0068)

84.75 (1.20)

86.13 (1.50)

58.480 (3.770)

  1. Averages are over 50 repetitions of randomly splitting the data into training (63 observations) and testing (31 observations). Reported average error rates are obtained from the test sets