Skip to main content

Advertisement

Table 20 Mispredictions rate and fit statistics for selected models

From: Ridge regression estimated linear probability model predictions of O-glycosylation in proteins with structural and sequence data

Model Mispredictions rate under a 50% cutoff probability Fit statistics
In the set of non-O-glycosylated sequences (Y=0), the percentage of those that have estimated probabilities of O-glycosylation greater than 50% (\( \hat{\mathrm{Y}}>0.5\Big) \) In the set of O-glycosylated sequences (Y=1), the percentage of those that have estimated probabilities of O-glycosylation less than or equal to 50% (\( \hat{\mathrm{Y}}\le 0.5\Big) \) KS Brier Score
Ordinary LS estimated LPM in Table 15 0.37 0.61 99.1% 0.009
RR estimated LPM (used for estimating the weights for the WLS estimated LPM in Table 11) 0.28 7.90 96.7% 0.084
LPM in Table 11 0.28 0.61 99.2% 0.009
LPLM with ρ = 0.82 in Table 19 0.83 3.55 96.6% 0.019