Skip to main content

Table 10 The results of different testing samples using XGBoost and GBDT after deleting the low-score Features

From: A universal Wi-Fi fingerprint localization method based on machine learning and sample differences

Model-test

GBDT

XGBoost

Floor success rate (%)

Point success rate (%)

Mean error (m)

Floor success rate (%)

Point success rate (%)

Mean error (m)

M0-T0

100.00

79.69

1.455

98.05

64.45

2.534

M0-T1

100.00

80.61

3.070

98.98

66.33

4.933

M0-T2

95.62

52.84

6.293

99.23

55.67

4.943

M1-T0

99.22

60.55

2.783

97.66

39.06

4.899

M1-T1

99.49

73.98

3.294

96.94

57.14

4.661

M1-T2

94.07

55.67

7.631

98.71

55.41

4.854

M2-T0

100.00

74.61

1.855

98.44

46.09

3.846

M2-T1

100.00

79.08

2.715

100.00

63.27

4.781

M2-T2

99.74

76.55

2.657

100.00

57.22

3.920

Mean

98.68

70.40

3.530

98.67

56.07

4.370