A County-Level Similarity Network of Electric Vehicle Adoption: Integrating Predictive Modeling and Graph Theory

TL;DR

The study addresses county-level variability in EV adoption by integrating predictive-model-based feature importances into a weighted Gower similarity framework, constructing a mutual $k$-NN network, and applying modularity-based clustering to identify 27 county clusters. It demonstrates both global patterns—such as declines in median income, education, and charging infrastructure with lower adoption—and local deviations where rurality or poverty alone do not fully explain adoption outcomes. The approach combines multiple predictive models, PERMANOVA validation, and Cohen's $d$-based cluster profiling to reveal distinct pathways to similar adoption levels, offering policy-makers cluster-specific insights for tailored interventions. Overall, the method provides a principled, interpretable way to translate county-contextual features into actionable, localized EV adoption strategies, while acknowledging data availability limitations and suggesting temporal extensions for future work.

Abstract

Electric vehicle (EV) adoption is essential for reducing carbon dioxide (CO2) emissions from internal combustion engine vehicles (ICEVs), which account for nearly half of transportation-related emissions in the United States. Yet regional EV adoption varies widely, and prior studies often overlook county-level heterogeneity by relying on broad state-level analyses or limited city samples. Such approaches risk masking local patterns and may lead to inaccurate or non-transferable policy recommendations. This study introduces a graph-theoretic framework that complements predictive modeling to better capture how county-level characteristics relate to EV adoption. Feature importances from multiple predictive models are averaged and used as weights within a weighted Gower similarity metric to construct a county similarity network. A mutual k-nearest-neighbors procedure and modularity-based community detection identify 27 clusters of counties with similar weighted feature profiles. EV adoption rates are then analyzed across clusters, and standardized effect sizes (Cohens d) highlight the most distinguishing features for each cluster. Findings reveal consistent global trends, such as declining median income, educational attainment, and charging-station availability across lower adoption tiers; while also uncovering important local variations that general trend or prediction analyses fail to capture. In particular, some low-adoption groups are rural but not economically disadvantaged, whereas others are urbanized yet experience high poverty rates, demonstrating that different mechanisms can lead to the same adoption outcome. By exposing both global structural patterns and localized deviations, this framework provides policymakers with actionable, cluster-specific insights for designing more effective and context-sensitive EV adoption strategies.

Figures (6)

• Figure 1: A descriptive caption for your figure.
• Figure 2: County-level similarity network based on socio-economic, political, environmental, and infrastructure features. Each node represents a county, and edges denote high similarity between counties across the standardized feature space.
• Figure 3: Pairwise cluster comparison heatmap showing the number of significantly different features between clusters. The color intensity represents the log-transformed p-values for better visualization of significance levels.
• Figure 4: Boxplots of EV adoption (EVs per 10,000 people) across the 27 clusters. Each box represents the distribution of EV adoption within a cluster.
• Figure 5: Standardized mean differences (Cohen’s d) for four negatively trending features—median income, charging stations per 10,000 population, high school graduation rate, and college completion rate—across clusters ordered by EV adoption rate. These structural features generally decline as EV adoption decreases, with minor variations between adjacent clusters.