Skip to main content
. 2017 Sep 1;5(3):197–212. doi: 10.1089/big.2017.0074

Table 3.

The effort to cloak different users' characteristics using a logistic regression with 100 singular-value decomposition components, a logistic regression, and Naive Bayes model

  Inline graphic Inline graphic
Task LRSVD LR NB LRSVD, % LR, % NB, %
Age Inline graphic37 10.3 7.3 37.7 7.7 7.4 17.9
Agreeableness Inline graphic5 5.0 2.9 7.2 2.3 4.3 12.6
Conscientiousness Inline graphic5 4.7 3.4 16.1 3.9 4.8 44.1
Extraversion Inline graphic5 4.4 3.6 58.0 1.9 2.5 10.2
IQ Inline graphic 90 6.9 3.7 21.6 4.5 7.3 7.2
IQ Inline graphic 130 6.6 2.9 14.4 2.8 3.3 9.4
Is democrat 8.5 9.4 61.7 1.7 2.0 10.6
Is drinking 6.8 5.4 17.1 2.0 2.1 8.2
Is female 10.0 11.6 377.4 1.9 2.0 25.9
Is gay 5.7 9.1 20.6 3.8 15.0 15.3
Is homosexual 3.5 3.4 8.2 2.4 3.9 10.8
Is lesbian 3.1 2.5 7.4 1.9 3.9 13.6
Is Muslim 11.7 8.9 31.1 9.6 10.1 46.5
Is single 13.7 10.2 105.8 3.4 2.8 12.5
Is smoking 8.4 7.0 26.2 2.8 3.2 13.5
Life satisfaction Inline graphic6 5.1 4.1 10.3 2.2 7.2 8.3
Network density Inline graphic65 10.5 2.6 75.7 2.1 3.9 7.7
Neuroticism Inline graphic5 9.1 2.3 254.5 2.2 3.6 18.0
Num friends Inline graphic585 5.0 4.7 52.6 2.1 2.5 10.6
Openness Inline graphic5 6.7 3.7 28.6 2.3 2.5 11.1
ss belief = 1 5.7 4.5 24.2 2.9 3.6 10.4
ss belief = 5 8.3 4.7 18.4 2.1 4.1 6.2
Uses drugs 12.2 8.2 31.5 2.7 3.4 9.0
Mean 7.5 5.5 56.8 3.1 4.6 14.8
Median 6.8 4.6 26.2 2.3 3.6 10.8

Absolute efforts are presented in the left panel, and relative efforts are in the right panel.

LR, logistic regression; LRSVD, logistic regression singular-value decomposition; NB, Naive Bayes model.