overview

Pipeline Overview

This Quarto documentation describes the analytical workflow used to generate results for:

Astle, W., & Haus, S. (2025)., Driving with Influence: Exploring Crash Factors of Automated Systems In Different Roadway Contexts

The analysis is managed using the targetspackage, which handles dependency tracking and reproducible builds.

Visualizing the Workflow

Diagrammatically, the workflow can be visualized as follows:

flowchart TD

    %% Define the look for subgraphs
    classDef blackText fill:none,color:black,font-weight:bold;

    %% Initial load (Blue)
    subgraph Initial_Datasets["Initial Datasets"]
        A[ADAS load]
        B[ADS load]
    end
    class Initial_Datasets blackText
    style A fill:#1f77b4,stroke:#000,stroke-width:2px,color:#fff
    style B fill:#1f77b4,stroke:#000,stroke-width:2px,color:#fff

    A --> C[Clean]
    B --> C
    style C fill:#2ca02c,stroke:#000,stroke-width:2px,color:#fff

    %% Make rules nodes (Bright Gold)
    C --> D[Make rules]
    style D fill:#ffcc00,stroke:#000,stroke-width:2px,color:#000

    %% Global Rules (Orange)
    subgraph Global_Rules["Global Rules"]
        E[ADAS rules]
        F[ADS rules]
    end
    class Global_Rules blackText
    style E fill:#ff7f0e,stroke:#000,stroke-width:2px,color:#000
    style F fill:#ff7f0e,stroke:#000,stroke-width:2px,color:#000

    G[Calculate influence]
    style G fill:#9467bd,stroke:#000,stroke-width:2px,color:#fff

    D --> E --> G
    D --> F --> G

    G --> H[ADAS Global scores]
    G --> I[ADS Global scores]
    style H fill:#d62728,stroke:#000,stroke-width:2px,color:#fff
    style I fill:#d62728,stroke:#000,stroke-width:2px,color:#fff

    %% Initial Influence Scores
    subgraph Global_Influence_Scores["Global Influence Scores"]
        H
        I
    end
    class Global_Influence_Scores blackText

    %% Join / Plot nodes (Cyan / Gray)
    H --> J[Join Global scores]
    I --> J
    style J fill:#17becf,stroke:#000,stroke-width:2px,color:#000

    J --> K[Plot Global scores]

    style K fill:#00bfae,stroke:#000,stroke-width:2px,color:#000

    %% Split step (Green)
    C --> N[Split]
    style N fill:#2ca02c,stroke:#000,stroke-width:2px,color:#fff

    subgraph Roadway_Datasets["Roadway Datasets"]
        O[ADAS Highway]
        P[ADAS Intersection]
        Q[ADS Highway]
        R[ADS Intersection]
    end
    class Roadway_Datasets blackText
    style O fill:#1f77b4,stroke:#000,stroke-width:2px,color:#fff
    style P fill:#1f77b4,stroke:#000,stroke-width:2px,color:#fff
    style Q fill:#1f77b4,stroke:#000,stroke-width:2px,color:#fff
    style R fill:#1f77b4,stroke:#000,stroke-width:2px,color:#fff

    %% Make rules nodes (Bright Gold)
    N --> O --> S[Make rules]
    N --> P --> S
    N --> Q --> S
    N --> R --> S
    style S fill:#ffcc00,stroke:#000,stroke-width:2px,color:#000

    %% Roadway Rules (Orange)
    S --> T[ADAS Highway Rules]
    S --> U[ADAS Intersection Rules]
    S --> V[ADS Highway Rules]
    S --> W[ADS Intersection Rules]
    style T fill:#ff7f0e,stroke:#000,stroke-width:2px,color:#000
    style U fill:#ff7f0e,stroke:#000,stroke-width:2px,color:#000
    style V fill:#ff7f0e,stroke:#000,stroke-width:2px,color:#000
    style W fill:#ff7f0e,stroke:#000,stroke-width:2px,color:#000

    subgraph Roadway_Rules["Roadway Rules"]
    T
    U
    V
    W
    end
    class Roadway_Rules blackText

    %% Roadway Rules Influence (Purple)
    T --> X[Calculate influence]
    U --> X
    V --> X
    W --> X
    style X fill:#9467bd,stroke:#000,stroke-width:2px,color:#fff

    %% Scores (Red)
    X --> Y[ADAS Highway scores]
    X --> Z[ADAS Intersection scores]
    X --> AA[ADS Highway scores]
    X --> AB[ADS Intersection scores]
    style Y fill:#d62728,stroke:#000,stroke-width:2px,color:#fff
    style Z fill:#d62728,stroke:#000,stroke-width:2px,color:#fff
    style AA fill:#d62728,stroke:#000,stroke-width:2px,color:#fff
    style AB fill:#d62728,stroke:#000,stroke-width:2px,color:#fff

    subgraph Roadway_Influence_Scores["Roadway Influence Scores"]
    Y
    Z
    AA
    AB
    end
    class Roadway_Influence_Scores blackText
    
    %% Join / Plot nodes (Cyan / Gray)
Z --> AC[Join Intersection scores] --> AD[Plot Int scores]
style AC fill:#17becf,stroke:#000,stroke-width:2px,color:#000
style AD fill:#00bfae,stroke:#000,stroke-width:2px,color:#000

Y --> AE[Join Highway scores] --> AF[Plot Hwy scores]
style AE fill:#17becf,stroke:#000,stroke-width:2px,color:#000
style AF fill:#00bfae,stroke:#000,stroke-width:2px,color:#000

AA --> AE
AB -->AC

How the Workflow Runs

The analysis is managed by the {targets} pipeline defined in _targets.R. This file outlines each step—data import, cleaning, analysis, and visualization—and their dependencies.

Custom functions live in the R/ folder and are automatically sourced by {targets}. Input data (public or simulated) are stored in data/, while processed outputs are created during the pipeline run.

Run the entire workflow with:

targets::tar_make()

You can then load the various outputs from the pipeline using:

targets::tar_load(target_name)

An Example

As an example, to view the cleaned data generated by the {targets} pipeline, I can:

1 - View the relevant part of the _targets.R file for reference.

In this instance, it would look like:

tar_target(
    cleaned_data, # this is the label of the target we would like to access
    clean_data(data),
    pattern = map(data),
    iteration = "list" # to access the two individual datasets, we have the output as a list
  )

2 - Run tar_make() to run the pipeline

targets::tar_make()

3 - Load the cleaned datasets using tar_load()

Notice that the same name from the tar_target portion of the code is used with tar_load.

targets::tar_load(cleaned_data)

4 - Access the dataset from the cleaned_data list

Since we had iteration = "list" in the _targets.R code, we can access each dataset separately from a list.

dataset_1 <- cleaned_data[[1]]
dataset_2 <- cleaned_data[[2]]

Pipeline Overview

Visualizing the Workflow

How the Workflow Runs

An Example

Further Reading