Efficient Histogram-Based Gradient Boosting Approach for Accident Severity Prediction With Multisource Data

Abstract

Many people lose their lives in road accidents because they do not receive timely treatment after the accident from emergency medical services; providing timely emergency services can decrease the fatality rate as well as the severity of accidents. In this study, we predicted the severity of car accidents for use by trauma centers and hospitals for emergency response management. The predictions of our model could be used to decide whether an ambulance unit should be dispatched to the crash site or not. This study used histogram-based gradient boosting (HistGBDT), a modification of the gradient boosting (GBDT) classifier that accelerates the learning process and increases a model’s prediction power. The HistGBDT model was compared with seven state-of-the-art machine learning models: logistic regression, multilayer perceptron, random forest, extremely randomized trees, bagging, AdaBoost, and GBDT. The experiments were conducted on French accident data from 2005 to 2018. The HistGBDT model, with an overall accuracy of 82.5%, recall of 76.7%, and precision of 81.9%, outperformed other models. An analysis of feature importance indicated that safety equipment was the most important feature and vehicle category, department, localization, and region were other significant features. The Fβ measure (i.e., the weighted harmonic mean of recall and precision) was optimized with different weights on recall for the four best performing models to compare the tradeoff between the two crucial performance measures.

Assessing the Influence of Sensor-Induced Noise on Machine-Learning-Based Changeover Detection in CNC Machines

“…The boosting family of algorithms is less susceptible to outliers, can handle a variety of categorical and numerical variables without considerable preprocessing, and is scalable with efficient and parallelized implementations for diverse classification problems [ 58 ]. The HISTGradientBoosting [ 59 ], LightGBM, Random Forest [ 60 ], DART [ 61 ], and XGBoost [ 62 ] algorithms were compared to find the best ML soft sensor that was most appropriate for this case study. A box plot, which is shown in Figure 4 , was used to reflect the probable outliers, central tendency, and dispersion for the HISTGradientBoosting, LightGBM, Random Forest, DART, and XGBoost algorithms in predicting changeovers in the presence of Gaussian noise.…”

Predicting Errors in Accident Hotspots and Investigating Spatiotemporal, Weather, and Behavioral Factors Using Interpretable Machine Learning: an Analysis of Telematics Big Data

“…To maximize data utilization, we used latitude, longitude, month, ambient light situation, hour, and other weather-related variables of each error to impute missing values in the weather conditions. Using histogram-based gradient boosting decision trees (HistGBDT) due to its ability to handle missing values in features [29][30][31] , with optimized performance by grid search, we achieved an accuracy of 93% and a weighted average F1-score of 92% in imputing missing values. Due to the large dataset and the study's objective of assessing all possible variables, we restricted our analysis to data without any missing values, resulting in a nal dataset comprising 619,988 rows.…”

“…29,618 (1.87% [ 95% CI: 1.85-1.89]) occurred in identi ed accident hotspots. Examining error types revealed that the majority of errors were associated with harsh turning (57.02% [56.95-57.10]) and fatigue (29.51%[29.44-29.58]).…”

Enhancing Assamese Word Recognition for CBIR: A Comparative Study of Ensemble Methods and Feature Extraction Techniques

“…Histogram-Based Gradient Boosting uses histograms to speed up training [24]. It optimizes the same loss function as traditional gradient boosting but employs histogram-based techniques for better efficiency.…”