Clint K. Morris

Clint's Resume

Link to Resume Close Window

Machine Learning

Understanding Multi-vehicle Collision Patterns on Freeways – A Machine Learning Approach

ABSTRACT: Generating meaningful inferences from crash data is vital to improving highway safety. Classic statistical methods are fundamental to crash data analysis and often regarded for their interpretability. However, given the complexity of crash mechanisms and associated heterogeneity, classic statistical methods, which lack versatility, might not be sufficient for granular crash analysis because of the high dimensional features involved in crash-related data. In contrast, machine learning approaches, which are more flexible in structure and capable of harnessing richer data sources available today, emerges as a suitable alternative. With the aid of new methods for model interpretation, the complex machine learning models, previously considered enigmatic, can be properly interpreted. In this study, two modern machine learning techniques, Linear Discriminate Analysis and eXtreme Gradient Boosting, were explored to classify three major types of multi-vehicle crashes (i.e., rear-end, same-direction sideswipe, and angle) occurred on Interstate 285 in Georgia. The study demonstrated the utility and versatility of modern machine learning methods in the context of crash analysis, particularly in understanding the potential features underlying different crash patterns on freeways. \n\n SKILLS: \n Web-scraping\n

Link to Journal Publication Close Window

Machine Learning

Improving Crash Classification Modeling with Resampling Methods

ABSTRACT: Crash data analysis is commonly subjected to imbalanced data. Varied by facility and control types, some crash types are more frequent than others. Uncommon crash types are routinely more severe and associated with higher economic and comprehensive costs and thus crucial to prevent. It is paramount to generate inferential models that can distinguish severe types of crashes from more common, lower risk events. The process of modeling towards infrequent instances induces data disparity. Therefore, mitigating and managing imbalanced data is essential to the development of meaningful inferences that help to reveal effective countermeasures. This study focuses on comparing the effects of data resampling on inferential machine learning and classical statistical models. These models were tasked with classifying different types of events observed in 5-minute windows across a day. The events include non-crash as a special case and four types of collision (i.e., rear-end, same-direction sideswipe, angle, single vehicle) on freeways. Among the five classes there exist a vast divergency in data representation, which necessitates resampling strategies to deal with the data imbalance. Specifically, the eXtreme Gradient Boosting, one of the most popular machine learning methods, and the classic nested logit model were compared. It was found that oversampling enhanced model performance on minor class prediction. The adaptive synthetic sampling approach achieved the best results, followed by random oversampling.

Link to Pending Research Close Window

America Society of Mechanical Engineers 2019

ABSTRACT: Currently there are no cost-effective ways to quantitatively measure the in-vivo mechanical properties of the Achilles tendon. Stiffness can be used as a measure of tone and mechanical integrity of both muscles and tendons. Stiffness of the Achilles tendon (AT) can be quantified by the speed of shear wave propagation. The speed of propagation can then be used to find the instantaneous shear modulus. Currently there are other methods such as Ultrasound (US) imaging and Magnetic Resonance Imaging (MRI) which are used clinically to determine the variations in stiffness of the AT. However, these methods require complex signal processing and experienced technicians. Moreover, US imaging technique is limited in measuring high shear wave speed values which are greater than 17 m/s. In this research, one-dimensional accelerometers were used to measure acceleration through the AT. Then a cross-correlation signal processing technique was used to convert acceleration to the velocity of shear wave propagation across the AT. This method could potentially evaluate the mechanical properties of both normal and damaged tendons. This process has proven to be a cost-effective and simple way to assess the stiffness of the AT.

Link to Publication Full Text Close Window

Deep Learning

GDOT project to merge machine learning and deep learning to replace Weigh-In-Motion (WIM) station with an optical alternative. This algorithm uses wheel spacing to ascertain truck/trailer type. In addition, identifying the presents of refrigerator units and automatically reverse image searching potential company logos.

Close Window

GitHub

Link to GitHub Close Window