Research Article

Keratoconus Severity Classification Using Features Selection and Machine Learning Algorithms

Table 5

Precision, recall, -score, accuracy, and execution time of different models using different features selection algorithms.

MethodModelKeratoconus classification using 2 classesKeratoconus classification using 4 classes
PrecisionRecall-scoreAccuracyTime in secondPrecisionRecall-scoreAccuracyTime in second

All featuresLR0.970.970.9797.38%1.3842560.930.940.9393.52%2.281342
LDA0.960.960.9696.17%1.9499490.90.90.989.88%1.747491
KNN0.960.960.9595.73%6.7927910.910.910.9191.37%5.829261
CART0.970.960.9696.14%5.5637090.920.920.9291.49%6.864419
NB0.950.930.9493.30%0.2730790.890.650.7164.80%0.242261
SVM0.930.920.8891.88%9.0764910.760.820.7582.36%14.471874
RF0.980.980.9898.00%16.0145120.950.950.9595.32%20.485403

Filter and mutual informationLR0.930.930.9393.46%0.8980350.810.870.8487.35%2.377594
LDA0.970.970.9796.93%0.2301030.870.90.8790.01%0.092326
KNN0.950.960.9595.61%0.2592090.830.870.8487.32%0.192763
CART0.960.960.9695.67%0.1800620.880.880.8887.83%0.190027
NB0.950.940.9494.06%0.0604320.870.80.8279.79%0.059485
SVM0.90.920.8991.97%0.7969060.770.830.7582.58%1.20752
RF0.970.970.9797.15%3.8846380.890.910.8990.54%4.793544

ANOVALR0.950.950.9595.29%1.032310.820.880.8588.49%1.463929
LDA0.960.960.9696.27%0.1440630.860.890.8689.28%0.092706
KNN0.950.960.9595.57%0.2034060.860.880.8787.92%0.186572
CART0.960.960.9695.61%0.1559270.870.870.8787.13%0.165356
NB0.950.930.9492.85%0.065920.880.690.7468.66%0.067247
SVM0.950.950.9595.35%0.6285570.80.860.8185.52%1.203339
RF0.970.970.9797.033.0673860.90.910.990.83%4.246253

EmbeddedLR0.970.970.9796.97%0.7724390.830.890.8588.87%1.47536
LDA0.970.970.9797.12%0.1832230.880.90.8789.66%0.191917
KNN0.960.960.9696.33%0.3510210.890.890.8888.74%0.198216
CART0.970.970.9796.55%0.3994750.880.880.8888.39%0.343518
NB0.960.950.9695.16%0.0591980.90.820.8582.16%0.061666
SVM0.960.960.9696.05%0.7452120.810.860.8286.24%1.343924
RF0.980.980.9897.63%4.7359320.910.910.9191.21%5.290628

Embedded and filterLR0.960.960.9696.08%0.716740.80.870.8386.62%1.191328
LDA0.960.960.9696.11%0.2163660.890.880.8588.49%0.059547
KNN0.960.960.9696.21%0.1855420.860.880.8688.43%0.151569
CART0.960.960.9695.45%0.0846780.860.860.8685.42%0.088294
NB0.940.940.9494.28%0.063540.880.890.8888.90%0.059538
SVM0.950.950.9494.88%0.5106190.80.860.8185.80%0.873448
RF0.970.970.9796.81%2.46460.890.90.8989.79%2.605076

Filter and RFELR0.980.980.9897.63%0.8021030.80.860.8286.15%2.173507
LDA0.970.970.9797.09%0.1803950.870.90.8789.91%0.091918
KNN0.950.960.9595.57%0.2870140.830.870.8487.29%0.16819
CART0.960.960.9696.21%0.1627090.90.890.8989.28%0.18652
NB0.950.940.9494.31%0.0644850.870.860.8786.31%0.055916
SVM0.90.920.8991.97%0.8382370.770.830.7582.58%1.209206
RF0.980.980.9897.63%3.7656890.930.930.9393.17%4.597132

RFELR0.940.940.9494.37%0.7244120.830.890.8588.87%1.428089
LDA0.970.970.9796.84%0.1738820.890.90.8789.88%0.09103
KNN0.970.970.9797.06%0.2722940.90.90.989.98%0.213538
CART0.970.970.9796.52%0.133980.910.90.9190.26%0.206897
NB0.960.960.9695.61%0.0559330.910.860.8785.52%0.05992
SVM0.960.960.9696.14%0.5029240.810.860.8286.43%1.160413
RF0.980.980.9897.91%3.2101880.940.940.9493.83%4.381234

Filter and HRFALR0.840.920.8891.65%0.2342210.70.830.7682.89%1.201195
LDA0.840.920.8891.65%0.0688740.70.830.7682.96%0.063042
KNN0.860.910.8890.99%0.1733180.740.810.7780.58%0.157148
CART0.860.870.8686.78%0.0786980.740.760.7576.13%0.075387
NB0.840.920.8891.65%0.0608430.750.830.7683.05%0.057565
SVM0.840.920.8891.65%0.5895990.70.830.7682.99%1.006607
RF0.860.890.8788.90%2.959560.740.780.7677.90%3.135792

Filter and SFSLR0.960.960.9696.30%0.6617030.810.870.8487.45%1.456912
LDA0.970.970.9796.96%0.0921820.870.890.8689.06%0.092238
KNN0.960.960.9695.86%0.204270.840.870.8586.75%0.198333
CART0.960.960.9695.98%0.1361620.90.90.989.98%0.226029
NB0.940.940.9494.21%0.0576390.890.860.8786.12%0.059771
SVM0.90.920.8991.97%0.6398690.810.870.8386.62%1.280597
RF0.980.980.9897.76%3.3770560.930.940.9393.64%4.575722

SFSLR0.980.980.9897.60%0.6278290.80.860.8286.34%1.447075
LDA0.970.970.9797.31%0.1504470.890.90.8889.95%0.091448
KNN0.960.960.9696.40%0.2642320.90.90.990.01%0.179493
CART0.970.970.9797.09%0.1485230.910.910.9191.33%0.158256
NB0.950.940.9494.12%0.0596450.910.860.8886.09%0.056018
SVM0.960.960.9595.89%0.5162740.760.850.885.04%1.071349
RF0.980.980.9898.10%3.2418810.950.950.9595.32%3.702065

GeneticLR0.970.980.9797.53%0.6609590.820.880.8487.98%1.426594
LDA0.970.970.9796.81%0.0905650.90.90.990.36%0.091082
KNN0.970.970.9796.81%0.1980130.840.870.8587.13%0.183919
CART0.970.970.9797.12%0.1489740.910.910.9190.20%0.191642
NB0.960.950.9594.75%0.0584660.910.880.8987.51%0.055448
SVM0.950.950.9595.32%0.6361810.810.870.8386.75%1.192417
RF0.980.980.9898.04%3.518570.940.940.9494.34%4.337631

Filter and SBSLR0.970.970.9796.93%0.6623590.810.870.8386.88%1.44051
LDA0.960.960.9696.24%0.0901070.880.90.8789.60%0.091316
KNN0.950.960.9595.67%0.1927130.840.870.8586.88%0.190318
CART0.960.960.9695.70%0.147320.90.90.990.26%0.189021
NB0.940.940.9494.44%0.0559370.890.830.8582.86%0.055396
SVM0.90.920.8991.94%0.6494430.810.870.8286.53%1.272575
RF0.970.970.9797.34%3.4321320.940.940.9393.71%4.532753

SBSLR0.980.980.9897.85%0.662359
LDA0.970.970.9797.31%0.090107
KNN0.960.960.9696.08%0.192713
CART0.970.970.9796.90%0.14732N.A
NB0.950.940.9493.87%0.055937
SVM0.950.950.9495.07%0.649443
RF0.980.980.9898.07%3.432132