Add Relationships Between Nodes
add_edges.RdAdd unweighted ties or weighted flows between nodes already in an atlas.
Usage
add_edges(x, data, from, to, weight = NULL, time = NULL,
layer = NULL, directed = FALSE,
coverage = c("edge_list", "observed_rows", "complete_panel"),
signed = FALSE, provenance = NULL)Arguments
- x
An atlas object.
- data
A data frame containing relationship endpoints and optional attributes.
- from
Column names containing relationship endpoint identifiers.
- to
Column names containing relationship endpoint identifiers.
- weight
Optional numeric weight column. Finite values and NA are supported. Negative values require signed = TRUE. NA and zero remain distinct.
- time
Optional column name for discrete, nonmissing time values. The viewer uses numeric-aware text sorting.
- layer
Optional column name for nonempty relationship types, such as Marriage or Business.
- directed
Whether relationships have a direction.
- coverage
An unweighted edge list, observed rows without inferred zeros, or a complete directed source-target-time-layer panel. The latter requires every combination except self-links, with explicit NA rows for unavailable observations. This contract is based on the IDs and times supplied.
- signed
Allow signed weights. Line color and dash encode sign. Arrow direction is preserved, circle area uses absolute incoming flows, and totals retain signs. Signed mode uses sign styling instead of source/layer styling.
- provenance
Optional column name for per-row reporting provenance, retained in the inspector, table, CSV, and JSON.
Details
Each layer can be added once. Nodes must precede relationships. Self-links, and duplicate relationships are rejected in this prototype. Signed weights require explicit opt-in. Multiple relationship types between the same endpoints are allowed. Distinct-partner degree uses enabled layers. Selection highlights incident relationships without redefining the degree population.
Weighted data require coverage = "observed_rows" or "complete_panel". For observed rows, an absent pair is not inferred to have a zero value. Complete panels require directed = TRUE and a time column. The expected roster is the Cartesian product of the distinct source IDs, target IDs, times and layers occurring in the supplied relationship table, excluding self-links. It is not inferred from all node IDs. Supply explicit NA rows for unavailable observations. Undirected duplicates include reversed pairs within the same time and layer.
Zero and missing weighted rows remain in data exports but draw no lines. Signed weights preserve arrow direction; negative lines are dashed amber and positive lines solid teal. Node circle areas use absolute incoming flows; totals retain signs. Source, year and enabled layers affect network totals; node focus highlights incident relationships.
Value
An offline_atlas object, suitable for piping into another
configuration function or export_html.
See also
add_nodes, set_controls, export_html.
For the tutorial, see the offlineatlas package overview.
Examples
families <- example_data("florence_nodes")
m <- atlas("Florentine families") |>
add_nodes(families, "id", "label", layout_x = "x", layout_y = "y",
coordinates = "schematic")
ties <- example_data("florence_edges")
m <- add_edges(m, ties, "from", "to", layer = "layer")
# An observed signed flow table, without inferred zeros.
places <- data.frame(id = c("Paris", "Rome"),
lon = c(2.35, 12.50), lat = c(48.86, 41.90))
m <- atlas("Places") |>
add_nodes(places, "id", longitude = "lon", latitude = "lat")
flows <- data.frame(from = "Paris", to = "Rome", amount = -5)
flows_map <- add_edges(m, flows, "from", "to", weight = "amount",
directed = TRUE, signed = TRUE,
coverage = "observed_rows")