knitr::opts_chunk$set(echo = FALSE, warning = FALSE, message = FALSE)

Introduction

Electric vehicles (EVs) have emerged as a transformative force in the global transition toward sustainability. Driven by advancements in clean energy technologies, evolving consumer preferences, and environmental concerns, EVs are reshaping the transportation landscape. Known for its ambitious “Clean Energy Transformation Act” plan by 2045, Washington State provides a unique setting to study the technical characteristics and adoption trends of EVs.

This project leverages a dataset sourced from Kaggle, containing information on Battery Electric Vehicles (BEVs) and Plug-in Hybrid Electric Vehicles (PHEVs) registered through the Washington State Department of Licensing (DOL). The dataset comprises approximately 133,658 observations and 17 variables, which include the following key attributes:

Key Dataset Variables

  • VIN..1.10.: Unique identifier for each vehicle
  • County: County of registration (e.g., King, Clark, Pierce)
  • City: City of registration
  • Model.Year: Model year of the vehicle
  • Make: Manufacturer (e.g., Tesla, Nissan, Chevrolet)
  • Model: Vehicle model (e.g., Model Y, Leaf, Bolt)
  • Electric.Vehicle.Type: Type of EV (Battery Electric Vehicle or Plug-in Hybrid Electric Vehicle)
  • CAFV.Eligibility: Clean fuel eligibility status
  • Electric.Range: Distance the vehicle can travel on a full charge, in miles
  • Base.MSRP: Manufacturer’s suggested retail price
  • Electric.Utility: Utility provider serving the vehicle’s charging location

For the purposes of this study and the sake of simplicity, we narrowed the dataset to vehicles registered in King, Clark, and Pierce counties, and excluded observations with missing electric range values. This focus ensures a detailed and manageable analysis of EV adoption in Washington State’s most populous regions.

In this project, we aim to investigate the technical characteristics and adoption patterns of EVs in Washington State by addressing the following research questions:

1. How is electric range influenced by EV types (BEVs and PHEVs) and manufacturers over time?

Hypothesis: BEVs, designed to operate fully on electric power, are expected to demonstrate higher ranges compared to PHEVs. Furthermore, advancements in battery technology and manufacturing practices should lead to an overall increase in electric ranges over time. Established manufacturers like Tesla, known for their innovation and state-of-the-art technologies, are anticipated to have a competitive edge with higher ranges.

2. What is the distribution of clean fuel-eligible EVs across Washington State?

Hypothesis: Given the increasing emphasis on clean energy, the majority of EVs are expected to qualify for clean fuel incentives. By exploring eligibility across counties like King, Clark, and Pierce, we can identify regional patterns and their potential connection to electric range.

3. How has Tesla’s model popularity evolved over time?

Hypothesis: Tesla’s Model Y, known for its affordability and versatility, is expected to dominate EV adoption in Washington State, followed by other models like the Model 3 and Model S. By analyzing temporal trends, we aim to highlight shifts in consumer preferences within Tesla’s product lineup.

By addressing these questions, this study aims to provide insights into the EV landscape in Washington State, exploring the interplay between technology and consumer behavior. The findings will inform relevant stakeholders, including policymakers, manufacturers, and consumers, about current trends and opportunities for growth in the EV market. This analysis will also highlight the role of EVs in advancing Washington’s clean energy goals and the broader global push for sustainable transportation.

Data Visualizations and Analysis

Research Quesion 1

Electric range is a critical feature of electric vehicles (EVs) that directly impacts their utility and attractiveness to consumers. It is defined as the distance an electric vehicle (EV) can travel on a single charge of its battery. This analysis examines how electric range varies by EV types—Battery Electric Vehicles (BEVs) and Plug-in Hybrid Electric Vehicles (PHEVs)—and explores whether advancements in technology or differences across manufacturers have influenced electric range over time.

To begin, a simple univariate histogram is used to visualize the distribution of electric range across EVs in the dataset.

The histogram shows that the majority of vehicles in the dataset have low electric ranges, predominantly between 0 and 50 miles. This indicates that a large number of EVs are likely Plug-in Hybrid Electric Vehicles (PHEVs) or models with limited electric capacity. Only a smaller fraction of vehicles exhibit ranges closer to 300-350 miles, which correlates with long-range Battery Electric Vehicles (BEVs) that usually associate with premium or advanced models.

This distribution suggests a potential disparity in EV capabilities, with most vehicles suited for shorter commutes. Fewer vehicles cater to consumers seeking extended travel on a single charge. Meanwhile, the lack of electric range data with many 0 values also poses challenges to further extrapolating from the grpah.

To further analyze how electric range varies over time and between EV types, a time-series scatter plot is employed. This provides insights into whether technological advancements or shifts in EV type preferences have influenced electric range over the years.

From the time-series graph, no significant trend in the electric range is observed for either BEV or PHEV types. While technological advancements in battery development have been a focus of the EV industry, these advancements are not clearly reflected in this dataset. For PHEVs, ranges remain largely consistent. For BEVs, the variability among data points may result from differences in market segmentation but remain relatively consistent as well.

The autocorrelation graph also indicates that electric range development might have plateaued in recent years or is influenced by other factors not captured in this dataset, such as technological barriers or cost considerations. For manufacturers, this stability in electric range suggests a possible equilibrium where battery improvements are used to enhance other vehicle attributes rather than further extending range.

Furthermore, the relationship between manufacturers and electric range can provide significant insights into the design philosophy and priorities of various EV makers. A word cloud visualization of the ‘Make’ variable is used to highlight the distribution of manufacturers in the dataset.

## <<DocumentTermMatrix (documents: 133658, terms: 47)>>
## Non-/sparse entries: 133760/6148166
## Sparsity           : 98%
## Maximal term length: 13
## Weighting          : term frequency (tf)

Given the skewed distribution of manufacturer representation, we limit subsequent analysis to Tesla, Nissan, Chevrolet, KIA, BMW, and Ford. This allows us to focus on manufacturers with sufficient data points to provide meaningful insights while simplifying the analysis.

By narrowing down to these key manufacturers, we can explore whether differences in electric range arise from variations in EV types (BEV vs. PHEV).

The graph above highlights how average electric range differs by manufacturers and vehicle types. Chevrolet appears as a standout for both BEVs and PHEVs, showcasing its strong emphasis on all-range performance. Tesla and Nissan also perform well in the BEV segment, reflecting their leadership in electric vehicle innovation. Among PHEVs, Chevrolet and BMW lead with relatively higher average ranges.

This analysis underscores how different manufacturers prioritize electric range based on their market focus and vehicle types.

To deepen the analysis of electric range among different manufacturers, a Welch Two-Sample t-test was performed to compare the mean electric ranges of Tesla and Nissan Battery Electric Vehicles (BEVs). This test evaluates whether there is a statistically significant difference in the average electric range between these two leading manufacturers.

## 
##  Welch Two Sample t-test
## 
## data:  tesla$Electric.Range and nissan$Electric.Range
## t = -12.37, df = 20232, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -10.703555  -7.775481
## sample estimates:
## mean of x mean of y 
##  67.10702  76.34654

\section*{Welch Two-Sample t-test Results}

The Welch Two-Sample t-test reveals a statistically significant difference in the average electric range of Tesla and Nissan Battery Electric Vehicles (BEVs), given by \(p < 2.2 \times 10^{-16}\). This allows us to reject the null hypothesis, which posited that there is no difference in the average electric range between Tesla and Nissan BEVs.

The 95% confidence interval for the difference in means, \([-10.70, -7.78]\), is entirely negative. This confirms that Nissan BEVs, on average, have a longer electric range than Tesla BEVs. Specifically, the results show that Tesla BEVs have an average electric range of approximately \(67.11\) miles, while Nissan BEVs average \(76.35\) miles, reflecting a difference of about \(9.24\) miles.

These findings shed light on the diverse strategies adopted by leading EV manufacturers to cater to distinct consumer needs. While Tesla dominates the market with its innovative technology and premium features, Nissan’s competitive edge in range optimization might offer a compelling alternative for certain consumer segments of the EV market.

Further analysis could explore whether these trends are consistent across different regions or global markets. Additionally, examining other variables such as price, battery efficiency, or charging speed might provide a more comprehensive understanding. These insights could help manufacturers refine their strategies to meet evolving consumer demands in the rapidly growing EV industry.

Research Quesiton 2

The second research question focuses on understanding the distribution of Clean Alternative Fuel Vehicle (CAFV) eligibility and how electric range and other numerical attributes, such as MSRP (Manufacturer’s Suggested Retail Price) and model year, relate to eligibility status. By exploring these relationships, we aim to identify trends and patterns in CAFV eligibility and its association with key attributes.

To simplify the complexity of the data while retaining its variability, Principal Component Analysis (PCA) was employed. This dimensionality reduction technique highlights the relationships between variables and identifies clusters within the data. The analysis was performed using scaled and log-transformed versions of three key variables: Electric Range, MSRP, and Model Year.

The PCA biplot offers a comprehensive view of the relationships among the key variables—Electric Range, Base MSRP, and Model Year—by reducing their complexity into two principal components. The first principal component (Dim1) explains 59% of the variance, while the second component (Dim2) accounts for 30.5%. Together, these components capture approximately 89.5% of the total variability, ensuring that the reduced-dimensional representation retains most of the dataset’s critical information.

It also shows three distinct clusters of EVs based on their electric range, base MSRP, and model year. The red cluster highlights premium, high-range vehicles that dominate clean-fuel eligibility. The blue cluster represents mid-range EVs that stabilize a balance between cost and performance. Finally, the green cluster includes older or budget-friendly vehicles with shorter ranges, reflecting limited eligibility for clean-fuel incentives.

The alignment of the variables in the biplot reveals distinct patterns. Base MSRP is strongly associated with Dim1, indicating that Base MSRP is a primary factor driving variability in the dataset. Similarly, Electric Range also aligns with Dim1, reflecting its positive correlation with Base MSRP. This suggests that higher-priced vehicles tend to have longer electric ranges. In contrast, Model Year aligns more closely with Dim2, highlighting its independence from the other two variables and reflecting the temporal aspect of EV development.

The clusters identified in the biplot further confirms these patterns. Vehicles in Cluster 1 are characterized by high electric ranges and premium pricing, representing advanced EV models that likely align with “Clean Alternative Fuel Vehicle Eligible” status due to their superior range capabilities. Cluster 2 contains mid-range vehicles that balance price and range, reflecting a moderate approach to EV design and affordability. Meanwhile, Cluster 3 includes older and lower-priced EVs with shorter ranges, likely aligning with “Not Eligible” or “Unknown” CAFV categories.

The analysis demonstrates that CAFV eligibility is intrinsically linked to these three variables. Higher eligibility appears to correlate with newer models that are more expensive and offer longer ranges. Conversely, vehicles with lower eligibility tend to be older or budget-friendly models with limited range capabilities. This reinforces the idea that technological advancements and premium features play a crucial role in determining a vehicle’s alignment with clean energy initiatives.

Electric Range and CAFV Eligibility:

To further examine the relationship between electric range and CAFV eligibility, a bar plot was created, displaying the distribution of vehicles across three eligibility categories—Eligible, Not Eligible, and Unknown—faceted by county (King, Clark, and Pierce).

The bar plot above illustrates the distribution of vehicles across three CAFV eligibility categories—Eligible, Not Eligible, and Unknown—faceted by county (King, Clark, and Pierce). The visualization reveals notable trends consistent across all three counties.

The “Unknown” eligibility category accounts for the majority of the data points and aligns predominantly with vehicles that have an electric range of 0 miles. This suggests that vehicles categorized as “Unknown” either lack electric capabilities or have incomplete data regarding their electric range, making it difficult to assess their eligibility for clean fuel incentives. This highlights a significant limitation in the dataset, as a substantial portion of the observations lack critical information.

The “Not Eligible” category corresponds primarily to vehicles with electric ranges between 1 and 50 miles. These vehicles likely include plug-in hybrid electric vehicles (PHEVs) with smaller batteries and fail to meet the range requirements for clean fuel incentives. On the other hand, vehicles in the “Eligible” category exhibit electric ranges exceeding 50 miles, aligning with Battery Electric Vehicles (BEVs) designed for extended travel distances. This pattern underscores the importance of electric range in determining CAFV eligibility, with longer ranges more likely to fulfill the criteria.

Interestingly, all three counties—King, Clark, and Pierce—show similar trends in the distribution of eligibility categories. This consistency suggests that CAFV eligibility patterns are largely influenced by vehicle characteristics rather than regional variations. However, it is also important to note the large proportion of vehicles in the “Unknown” category, which could obscure potential regional differences in eligibility.

To conclude, this analysis underscores the critical role of electric range in CAFV eligibility determination and highlights the challenges posed by missing or incomplete data. Future work could focus on addressing these gaps by obtaining additional information on vehicles in the “Unknown” category.

Research Question 3

As the leading EV manufacturer, Tesla has introduced multiple innovative and cutting-edge models to the market, such as the Model S, Model 3, Model X, and Model Y. To understand Tesla’s impact and the evolution of its product line, this research question investigates the distribution of Tesla model types and provides insights into consumer preferences for Tesla vehicles.

A univariate data analysis was performed on Tesla’s model types, leveraging error bars to visualize confidence intervals for the proportion of each model in the dataset. This approach helps us assess the relative popularity of different Tesla models while accounting for variability in the data.

##           CYBERTRUCK   MODEL 3    MODEL S    MODEL X   MODEL Y     ROADSTER
## CI.lower 0.005368972 0.3547237 0.08259471 0.07004914 0.4741242 0.0003667105
## CI.upper 0.006586843 0.3623003 0.08699594 0.07413540 0.4820161 0.0007379028

The dominance of Tesla in the EV market is reflected through the popularity of its various models. In this research question, we explore two key aspects: the distribution of Tesla’s model types and how their popularity has evolved over time in Washington State.

The bar plot with confidence intervals highlights the distribution of Tesla model counts among WA EV buyers. Model Y is overwhelmingly the most popular Tesla model, significantly outpacing all other models, followed by Model 3. This aligns with Tesla’s strategic focus on delivering mass-market appeal through the versatility of Model Y and the affordability of Model 3.

The lack of overlap in the error bars indicates statistically significant differences in model counts. This provides strong evidence that WA buyers have clear preferences when it comes to Tesla’s lineup. Models S and X, while still present, account for much smaller proportions of the dataset, likely due to their higher price points and more niche market appeal. The Cybertruck and Roadster represent even smaller segments, as these newer models cater to specific buyer demographics and are less widely available.

To further investigate the dynamics of Tesla’s market presence, the distribution of Tesla models by model year is included

The graph illustrates Tesla model adoption trends over time in Washington State. Model Y demonstrates the fastest growth rate, particularly between 2022 and 2023, highlighting its increasing popularity among EV buyers. This surge can likely be attributed to the Model Y’s combination of affordability, practicality, and Tesla’s advanced technology offerings.

In contrast, the adoption rates of Model S, Model 3, and Model X remain relatively stable across the years. While these models continue to contribute to Tesla’s market presence, their slower growth suggests that they have reached a more mature phase in their adoption lifecycle. The Model 3, despite being one of Tesla’s most affordable models, appears to be overshadowed by the increasing appeal of the Model Y.

The Cybertruck and Roadster show minimal adoption during the observed years, reflecting their niche market appeal and limited production or availability during this time frame. The Cybertruck adoption has not yet reached significant levels in Washington State.

This analysis highlights Tesla’s ability to meet diverse market needs through its product lineup. The rapid adoption of the Model Y underscores a broader consumer preference for EVs that balance affordability and utility, while the steady presence of other models reflects their ongoing appeal to specific buyer segments. Future analyses could explore how these trends vary across different geographic regions or correlate with broader economic and policy shifts, such as tax incentives.

Conclusion

The findings of this project provide valuable insights into the evolving electric vehicle (EV) landscape in Washington State. Through detailed analyses and visualizations, we explored the technical characteristics and adoption patterns of EVs, addressing the three research questions effectively while drawing meaningful conclusions.

For Research Question 1, we examined how electric range varies across EV types and manufacturers over time. As hypothesized, Battery Electric Vehicles (BEVs) demonstrate significantly higher electric ranges compared to Plug-in Hybrid Electric Vehicles (PHEVs). However, contrary to expectations, the time-series and autocorrelation analyses revealed no observable trend in electric range improvements over the period from 2008 to 2024. This muted change suggests that advancements in battery technology have not yet resulted in substantial increases in electric range, pointing to the long development cycle of EV batteries. The slow pace of breakthroughs in battery energy density and efficiency underscores the challenges facing the industry in achieving significant progress in increasing electric range.

Research Question 2 focused on the distribution of clean-fuel eligibility across counties and its relationship with electric range. While our analysis confirmed that vehicles with higher electric ranges are more likely to meet clean-fuel eligibility requirements, a surprising finding emerged: a significant portion of the dataset falls into the “Unknown” eligibility category, particularly for vehicles with 0 electric range. This data gap presents a challenge to drawing definitive conclusions about eligibility patterns. Nonetheless, the PCA and bar plots revealed a clear relationship between electric range, suggested price, and model year, with higher eligibility linked to newer, premium models offering extended ranges. This insight underscores the importance of electric range as a key determinant of clean-fuel eligibility.

For Research Question 3, we investigated Tesla’s model popularity over time. The analysis confirmed our hypothesis that the Model Y dominates the market, experiencing the fastest growth in adoption, particularly between 2022 and 2023. Other models, such as the Model 3, maintain steady adoption rates but have not matched the rapid rise of the Model Y. Meanwhile, tailored models like the Cybertruck and Roadster represent smaller market segments, likely due to their specialized appeal and limited availability. These findings underscore Tesla’s ability to cater to diverse consumer needs and solidify its leadership position in the EV market.

In summary, this study highlights the complex interplay of factors influencing EV adoption in Washington State, from range and price to eligibility and manufacturer strategies.

Discussion

Several limitations and unanswered questions point to opportunities for future work. A notable challenge encountered in this study is the prevalence of missing or unknown values in the dataset, particularly for the “CAFV Eligibility” variable. This significantly limits the ability to fully analyze clean-fuel eligibility patterns and their relationship with other variables such as electric range and MSRP. Future research could address this by acquiring more complete and reliable datasets, potentially incorporating updated or supplementary information from manufacturers, state records, or other external sources.

Additionally, our analysis focuses predominantly on technical characteristics such as electric range, manufacturers, and model types, and they do not fully capture the range of factors influencing consumer preferences and purchase decisions. Variables such as price elasticity, availability of charging infrastructure, geographic variations in charging station density, and government incentives are critical to providing a holistic view of the EV market. These factors, which were beyond the scope of this project, could be integrated into future studies to create a more comprehensive investigation of EV adoption dynamics.

Finally, this project relied on straightforward statistical techniques and visualizations to derive insights from the dataset. We recommend future researchers to focus on more advanced techniques such as machine learning, clustering algorithms, or econometric modeling could uncover deeper patterns such as forecasting future EV trends and identifying distinct consumer segments demographically. The inclusion of such techniques would provide more insightful conclusions for both policymakers and manufacturers.