Page 128 - Data Science Algorithms in a Week

112 Loris Nanni, Sheryl Brahnam and Alessandra Lumini

Table 2: Performance of each reshaping method coupled with the different
texture descriptors for each dataset

LPQ Tr CW RS DCT FFT
DATASET AUC AUC AUC AUC AUC

breast 98.0 97.6 0 96.8 97.3
heart 64.0 90.4 0 86.9 88.0
pima 53.1 73.6 0 71.6 71.4

sonar 60.9 92.6 92.1 93.0 93.6
iono 86.2 98.8 98.5 98.6 98.3
liver 56.7 68.9 0 70.8 71.6
hab 48.7 0 0 63.5 63.4

vote 49.1 96.9 0 97.7 97.9
aust 71.7 91.2 0 90.1 90.5
trans 52.4 0 0 68.0 67.6
wdbc 89.9 97.9 97.7 98.6 98.9

bCI 76.7 96.2 93.4 96.6 96.7
pap 70.3 84.2 85.7 87.2 88.1
torn 80.2 89.3 93.3 93.6 93.6
gCr 72.6 73.5 77.6 78.2 78.3
Average 68.7 76.7 42.5 86.1 86.3

CLBP Tr CW RS DCT FFT
DATASET AUC AUC AUC AUC AUC

breast 98.5 97.4 98.2 97.1 97.7
heart 74.3 90.3 89.9 88.1 88.2
pima 60.3 73.2 70.8 71.5 72.0

sonar 65.6 90.5 90.1 91.5 92.7
iono 86.8 96.2 98.1 98.6 98.4
liver 56.9 68.8 70.7 70.7 68.5
hab 59.6 60.0 0 63.3 64.2
vote 50.1 96.1 97.6 96.9 97.4

aust 74.8 91.0 91.3 90.7 90.9
trans 65.8 64.5 69.4 66.1 67.8