Fig 1.
Framework scheme of clustering-based driving style classification method.
Fig 2.
Framework scheme of driving style classification method based on clustering and classification.
Fig 3.
Driving style classification model flow based on ensemble learning.
Table 1.
The details of the dataset.
Table 2.
Data attributes and their meanings.
Table 3.
Data preprocessing procedure.
Table 4.
Driving behavior feature parameters and their units.
Table 5.
Pre-classification procedures.
Table 6.
Classification procedures.
Table 7.
Experimental configuration.
Fig 4.
(a) Display of the first 10,000 speed data of the vehicle; (b) Display of the first 10,000 direction angle data of the vehicle; (c) Display of the vehicle mileage data.
Fig 5.
The result of micro-trip division of car AA00002.
Table 8.
The results of feature parameters extraction.
Table 9.
Clustering results.
Fig 6.
Classification results of test set sample by utilizing the proposed model.
(a) Driving style recognition result with accuracy = 98.85%; (b) Driving style recognition result with accuracy = 100%.
Fig 7.
Classification results of unclassified samples.
Table 10.
Classification results of unclassified samples.
Fig 8.
Visualization of clustering and classification results.
Fig 9.
Driving style statistics results.
Table 11.
Comparison of the values of DBI and CH indicator.
Fig 10.
Model evaluation indicators comparison.
(a) The comparison of model accuracy; (b) The comparison of model precision; (c) The comparison of model recall; (d) The comparison of model F-measure.
Table 12.
Comparison of the average value of evaluation indicators of different ensemble learning models.
Table 13.
Comparison of the value of evaluation indicators of different models for the same dataset.