Skip to main content

Table 10 Classification statistics for logit model compared to simulations

From: Manual wheelchair downhill stability: an analysis of factors affecting tip probability

 

Tip category

Forward tip

Backward tip

Rolled over

Stopped

Prevalence

0.270

0.106

0.547

0.076

Sensitivity

0.928

0.540

0.949

0.921

Specificity

0.977

0.954

0.869

0.993

PPV

0.939

0.663

0.912

0.914

NPV

0.973

0.933

0.913

0.994

F1 score

0.934

0.595

0.930

0.917

  1. Overall there was a 10.2% misclassification rate when comparing the predicted result from the multinomial logistic analysis to the simulation results. Categories ‘forward tip’, ‘rolled over’, and ‘stopped’ all had F1 scores over 0.9, and ‘backward tip’ was the least accurate category with an F1 score of 0.595