Convolute a spatiotemporal network to perform TNKDE

This function performs Temporal Network Kernel Density Estimation (TNKDE) on a spatiotemporal network object. It calculates density values across the network's segments for a series of time points.

Usage

convolute_spatiotemporal_network(
  segmented_network,
  kernel_space = compute_epanechnikov,
  kernel_time = compute_epanechnikov,
  bandwidth_space = 3,
  bandwidth_time = 2,
  time_points = 0:23,
  use_esd = TRUE,
  correct_boundary_effects = TRUE
)

Arguments

segmented_network

A spatiotemporal network object, typically the output of create_spatiotemporal_network() followed by set_events().

kernel_space

The spatial kernel function (default: compute_epanechnikov).

kernel_time

The temporal kernel function (default: compute_epanechnikov).

bandwidth_space

The spatial bandwidth (in the network's distance units).

bandwidth_time

The temporal bandwidth (in hours).

time_points

A numeric vector of time points (e.g., hours 0:23) at which to estimate the density.

use_esd

If TRUE (default), uses the Equal Split Discontinuous (ESD) kernel for spatial smoothing to account for network intersections.

correct_boundary_effects

If TRUE (default), corrects for boundary effects by normalizing the kernel weights from each source to sum to 1.

Value

The input network object with the $segments component updated to an sf data frame of spatial segments, now containing new columns (density_t_0, density_t_1, etc.) with the calculated density values. The total density over all segments and time points is normalized to sum to 1.

Examples

if (FALSE) { # \dontrun{
# Create a network and assign events
tnkde_input <- sample_roads |>
  create_road_network() |>
  create_spatiotemporal_network(
    spatial_length = 0.5,
    temporal_length = "1 hour"
  ) |>
  set_events(sample_accidents, time_column = "time")

# Run the TNKDE calculation
tnkde_result <- convolute_spatiotemporal_network(
  tnkde_input,
  bandwidth_space = 2,
  bandwidth_time = 1.5,
  time_points = 0:23
)

# View the results
print(tnkde_result$segments)

# Plot the results
# Plot a 2D snapshot for a specific time
plot(tnkde_result, mode = "density", snapshot_time = 19)

# Plot the full 3D space-time cube
plot(tnkde_result, mode = "density", plot_3d = TRUE)
} # }