Computes comprehensive metrics for evaluating clustering quality, including silhouette analysis, network modularity, and information-theoretic measures.
Usage
compute_tuning_metrics(
segments = NULL,
weight_matrix = NULL,
include_silhouette = TRUE,
include_modularity = TRUE
)Arguments
- segments
MONECA segmentation results (optional, for full analysis).
- weight_matrix
Weight matrix from mobility data (required).
- include_silhouette
Logical indicating whether to compute silhouette coefficients. Default is TRUE.
- include_modularity
Logical indicating whether to compute network modularity. Default is TRUE.
Value
List containing:
- silhouette
Average silhouette coefficient
- modularity
Network modularity score
- network_density
Network density measure
- connectivity
Network connectivity metrics
- overall
Overall quality score combining all metrics
Details
The function evaluates clustering quality using multiple complementary metrics:
Silhouette Analysis: Measures how well-separated clusters are by comparing within-cluster and between-cluster distances.
Network Modularity: Assesses community structure quality in the mobility network.
Connectivity Metrics: Evaluates network structure properties including density, component structure, and edge distribution.
Examples
if (FALSE) { # \dontrun{
# Generate sample data and create weight matrix
mobility_data <- generate_mobility_data(n_classes = 5, seed = 123)
weight_matrix <- weight.matrix(mobility_data, small.cell.reduction = 2)
# Compute quality metrics
quality <- compute_tuning_metrics(weight_matrix = weight_matrix)
print(quality)
} # }