Fig 1.
The framework of ARM & RGS method.
Table 1.
Descriptive statistics of crash data.
Fig 2.
Modular RGS based on different graph properties.
(a) In-degree, (b) Out-degree, (c) Betweenness, and (d) PageRank.
Table 2.
Contributions of factors to the graph structures based on closeness.
Fig 3.
Group matrix-based visualization of 1452 rules.
Fig 4.
Distribution of lift, support, and confidence.
Table 3.
Rules sorted by top indicator (lift, support, and confidence).
Fig 5.
Graph structure for specific rule set.
(a) Severity: A & Severity: B, (b) Mountainous & Plain (terrain), and (c) Freeway & 2nd-class highway (road category).
Table 4.
Rules involving high-frequency RHS items.
Fig 6.
Contributing factors of multi-fatality crashes and normal crashes.