Automatic identification of bird females using egg phenotype Author Šulc, Michal Author Hughes, Anna E Author Troscianko, Jolyon Author Štětková, Gabriela Author Procházka, Petr Author Požgayová, Milica Author Piálek, Lubomír Author Piálková, Radka Author Brlík, Vojtěch Author Honza, Marcel text Zoological Journal of the Linnean Society 2022 2022-05-01 195 1 33 44 https://academic.oup.com/zoolinnean/article/195/1/33/6357656 journal article 10.1093/zoolinnean/zlab051 0024-4082 6530631

Table 2. The importance of individual variables in egg clustering using random forest analysis

Variable	Mean decrease in accuracy	Main PCA loadings
Variables with larger mean decrease in accuracy are more important for classifying the data (mean decrease in accuracy is a measure of how much the accuracy of the random forest decreases due to the exclusion/permutation of a single variable).The main PCA loadings are those that were greater than +/– 0.25.
PC2_pattern	28.42	Skew values at pattern energy scales 1, 0.707, 0.5, 0.3536, 0.25,
		0.1768, 0.125, 0.08839, 0.0625, 0.04419
PC2_spectra	26.80	426, 447, 468, 489, 510, 531nm
PC3_shape	23.81	Length, max width
PC1_shape	21.37	Length, max width, volume, surface area
PC1_spectra	19.79	342, 552, 573, 594, 636, 678, 699nm
UV_shape	19.36	-
PC2_shape	16.91	Ellipse deviation
PC1_luminance	15.42	Luminance sections 1, 2 and 3, standard deviation sections 1, 2
		and 3
PC3_pattern	15.18	Pattern energy scales 1, 0.7071, 0.5, 0.3536, 0.04419, 0.03125,
		0.0221
Brightness	12.90	-
PC1_pattern	11.23	Pattern energy scales 0.3536, 0.25, 0.1768, 0.125, 0.08839,
		0.0625, total pattern energy, total pattern energy in segment 2