Table 9 The results of different testing samples using 23 features with XGBoost and GBDT
From: A universal Wi-Fi fingerprint localization method based on machine learning and sample differences
Model-test | XGBoost | GBDT | ||||
|---|---|---|---|---|---|---|
Floor success rate (%) | Point success rate (%) | Mean error (m) | Floor success rate (%) | Point success rate (%) | Mean error (m) | |
M0-T0 | 98.83 | 60.94 | 2.527 | 100.00 | 81.25 | 1.300 |
M0-T1 | 98.98 | 68.37 | 4.311 | 98.98 | 79.08 | 3.231 |
M0-T2 | 99.23 | 59.28 | 4.753 | 96.65 | 59.02 | 7.421 |
M1-T0 | 98.83 | 57.81 | 3.104 | 99.61 | 69.14 | 2.363 |
M1-T1 | 99.49 | 62.76 | 4.861 | 99.49 | 78.06 | 2.535 |
M1-T2 | 99.23 | 47.42 | 5.835 | 83.25 | 44.85 | 9.109 |
M2-T0 | 97.66 | 48.83 | 4.241 | 100.00 | 75.00 | 2.035 |
M2-T1 | 100.00 | 59.69 | 5.241 | 100.00 | 80.10 | 2.757 |
M2-T2 | 99.49 | 45.88 | 5.227 | 97.68 | 75.26 | 3.032 |
Mean | 99.08 | 56.77 | 4.460 | 97.30 | 71.31 | 3.750 |