Data-Driven Insights Into EV Charging Patterns: Machine Learning Models Reveal Key Predictors of Station Utilization in Tennessee
This study analyzes electric vehicle (EV) charging patterns and station utilization in Tennessee using machine learning (ML) techniques. While previous research has examined time series usage data, few studies have incorporated point of interest (POI) information or explored the relationship between...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10935329/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850207284002553856 |
|---|---|
| author | Seyedmehdi Khaleghian Thanh-Nam Doan Joe Knox Austin Harris Mina Sartipi |
| author_facet | Seyedmehdi Khaleghian Thanh-Nam Doan Joe Knox Austin Harris Mina Sartipi |
| author_sort | Seyedmehdi Khaleghian |
| collection | DOAJ |
| description | This study analyzes electric vehicle (EV) charging patterns and station utilization in Tennessee using machine learning (ML) techniques. While previous research has examined time series usage data, few studies have incorporated point of interest (POI) information or explored the relationship between various features and charging station (CS) utilization, particularly in the context of U.S. public charging infrastructure. Moreover, existing studies have often relied on limited feature sets and achieved relatively low coefficients of determination (R-squared). To address these gaps, we utilize a dataset consisting of 49,900 charging session records of Level 2 (L2) and direct current fast charging (DCFC) stations across Tennessee from 2018 to June 2024. To the best of our knowledge, our work is the first study which extensively analyzes utilization under various features, especially POI info. By employing a range of supervised ML models, including Ordinary Least Squares (OLS) regression, Gradient Boosting, XGBoost, Multi-layer perceptron (MLP), and Random Forest (RF), we predict daily charging station utilization with significantly improved accuracy. Our models achieve a 74% improvement in R-squared values for L2 stations and a 72% improvement for DCFC stations compared to the state-of-the-art research baseline. These results demonstrate that our approach outperforms the models used in previous studies. |
| format | Article |
| id | doaj-art-ef260950c1d049d78f48c3ffcb7ec50e |
| institution | OA Journals |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-ef260950c1d049d78f48c3ffcb7ec50e2025-08-20T02:10:34ZengIEEEIEEE Access2169-35362025-01-0113514585148310.1109/ACCESS.2025.355308610935329Data-Driven Insights Into EV Charging Patterns: Machine Learning Models Reveal Key Predictors of Station Utilization in TennesseeSeyedmehdi Khaleghian0https://orcid.org/0000-0002-8578-3352Thanh-Nam Doan1Joe Knox2Austin Harris3https://orcid.org/0009-0002-5449-7174Mina Sartipi4https://orcid.org/0000-0002-6709-5046Center for Urban Informatics and Progress (CUIP), The University of Tennessee at Chattanooga, Chattanooga, TN, USACenter for Urban Informatics and Progress (CUIP), The University of Tennessee at Chattanooga, Chattanooga, TN, USASeven States Power Corporation, Chattanooga, TN, USACenter for Urban Informatics and Progress (CUIP), The University of Tennessee at Chattanooga, Chattanooga, TN, USACenter for Urban Informatics and Progress (CUIP), The University of Tennessee at Chattanooga, Chattanooga, TN, USAThis study analyzes electric vehicle (EV) charging patterns and station utilization in Tennessee using machine learning (ML) techniques. While previous research has examined time series usage data, few studies have incorporated point of interest (POI) information or explored the relationship between various features and charging station (CS) utilization, particularly in the context of U.S. public charging infrastructure. Moreover, existing studies have often relied on limited feature sets and achieved relatively low coefficients of determination (R-squared). To address these gaps, we utilize a dataset consisting of 49,900 charging session records of Level 2 (L2) and direct current fast charging (DCFC) stations across Tennessee from 2018 to June 2024. To the best of our knowledge, our work is the first study which extensively analyzes utilization under various features, especially POI info. By employing a range of supervised ML models, including Ordinary Least Squares (OLS) regression, Gradient Boosting, XGBoost, Multi-layer perceptron (MLP), and Random Forest (RF), we predict daily charging station utilization with significantly improved accuracy. Our models achieve a 74% improvement in R-squared values for L2 stations and a 72% improvement for DCFC stations compared to the state-of-the-art research baseline. These results demonstrate that our approach outperforms the models used in previous studies.https://ieeexplore.ieee.org/document/10935329/Charging stationelectric vehiclemachine learningstation utilizationdata-driven forecast |
| spellingShingle | Seyedmehdi Khaleghian Thanh-Nam Doan Joe Knox Austin Harris Mina Sartipi Data-Driven Insights Into EV Charging Patterns: Machine Learning Models Reveal Key Predictors of Station Utilization in Tennessee IEEE Access Charging station electric vehicle machine learning station utilization data-driven forecast |
| title | Data-Driven Insights Into EV Charging Patterns: Machine Learning Models Reveal Key Predictors of Station Utilization in Tennessee |
| title_full | Data-Driven Insights Into EV Charging Patterns: Machine Learning Models Reveal Key Predictors of Station Utilization in Tennessee |
| title_fullStr | Data-Driven Insights Into EV Charging Patterns: Machine Learning Models Reveal Key Predictors of Station Utilization in Tennessee |
| title_full_unstemmed | Data-Driven Insights Into EV Charging Patterns: Machine Learning Models Reveal Key Predictors of Station Utilization in Tennessee |
| title_short | Data-Driven Insights Into EV Charging Patterns: Machine Learning Models Reveal Key Predictors of Station Utilization in Tennessee |
| title_sort | data driven insights into ev charging patterns machine learning models reveal key predictors of station utilization in tennessee |
| topic | Charging station electric vehicle machine learning station utilization data-driven forecast |
| url | https://ieeexplore.ieee.org/document/10935329/ |
| work_keys_str_mv | AT seyedmehdikhaleghian datadriveninsightsintoevchargingpatternsmachinelearningmodelsrevealkeypredictorsofstationutilizationintennessee AT thanhnamdoan datadriveninsightsintoevchargingpatternsmachinelearningmodelsrevealkeypredictorsofstationutilizationintennessee AT joeknox datadriveninsightsintoevchargingpatternsmachinelearningmodelsrevealkeypredictorsofstationutilizationintennessee AT austinharris datadriveninsightsintoevchargingpatternsmachinelearningmodelsrevealkeypredictorsofstationutilizationintennessee AT minasartipi datadriveninsightsintoevchargingpatternsmachinelearningmodelsrevealkeypredictorsofstationutilizationintennessee |