Skip to main content

Table 4 Classification accuracy of machine learning algorithms compared with BioCAT

From: An image database of Drosophila melanogaster wings for phenomic and biometric analysis

Classification

Algorithm

Hessian

Shape

Shape + Size

Sex

Random forest (10)

85.0 %

92.3 % (±3.7)

94.7 % (±2.6)

 

Random forest (1,000)

85.0 %

96.1 % (±2.2)

95.9 % (±2.1)

 

SVM

81.7 %

99.0 % (±1.2)

99.0 % (±1.2)

Genotype

Random forest (10)

52.0 %

43.3 % (±3.5)

44.7 % (±3.7)

 

Random forest (1,000)

46.7 %

69.1 % (±3.4)

70.2 % (±2.8)

 

SVM

43.3 %

75.1 % (±2.8)

75.8 % (±2.7)

  1. Hessian column represents accuracy of classifications based on Hessian features extracted with BioCAT. Shape column represents classification accuracy based on landmarks and semi-landmarks, not including centroid. Shape + size represents classification accuracy based on landmarks and semi-landmarks, including centroid