Visualization Guide

AgenticWaves provides a comprehensive suite of publication-quality visualization functions for financial network analysis, agent-based modeling, and spillover detection.

Overview

The visualization framework includes:

Network visualizations using ggraph for publication quality
Time series plots with regime highlighting
Heatmaps for spillover matrices and correlations
Dashboard-style layouts for comprehensive analysis
Interactive plots with plotly integration
Agent visualization showing behavioral clustering

Network Visualizations

Enhanced Network Plots

library(AgenticWaves)

# Create sample network data
data <- get_sample_data("global_markets", n_assets = 10, n_periods = 500)
agents <- create_enhanced_agent_population(n_agents = 300)

sim_results <- simulate_enhanced_market_dynamics(
  agents = agents,
  asset_data = data,
  n_periods = 1000
)

spillover_results <- calculate_dynamic_spillover_networks(
  simulation_results = sim_results,
  window_size = 100
)

# Create enhanced network plot
network_plot <- plot_enhanced_network(
  spillover_matrix = spillover_results$average_spillover_matrix,
  layout = "stress",
  node_size_var = "degree",
  color_scheme = "viridis",
  title = "Financial Network - Average Spillovers"
)

print(network_plot)

Layout Options

Different network layouts reveal different structural aspects:

# Stress layout - good for general visualization
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  layout = "stress"
)

# Fruchterman-Reingold - emphasizes clusters
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  layout = "fr"
)

# Kamada-Kawai - good for small networks
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  layout = "kk"
)

# Circular layout - symmetric presentation
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  layout = "circle"
)

Color Schemes

# Viridis (default) - perceptually uniform
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  color_scheme = "viridis"
)

# Plasma - high contrast
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  color_scheme = "plasma"
)

# RdYlBu - diverging colors
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  color_scheme = "RdYlBu"
)

# Spectral - rainbow colors
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  color_scheme = "Spectral"
)

Agent-Based Visualizations

Agent Type Networks

# Create agent network
network <- create_dynamic_multilayer_network(
  agents = agents,
  network_types = c("trading", "information", "social"),
  density = 0.1
)

# Visualize agent network by type
agent_plot <- plot_agent_type_network(
  agents = agents,
  network_obj = network$networks$trading$graph,
  layout = "stress"
)

print(agent_plot)

Wealth Distribution

# Plot agent wealth evolution
library(ggplot2)

wealth_data <- data.frame(
  time = rep(1:nrow(sim_results$agent_wealth), ncol(sim_results$agent_wealth)),
  agent_id = rep(1:ncol(sim_results$agent_wealth), each = nrow(sim_results$agent_wealth)),
  wealth = as.vector(sim_results$agent_wealth),
  agent_type = rep(sapply(agents, function(x) x$type), each = nrow(sim_results$agent_wealth))
)

# Wealth evolution by type
ggplot(wealth_data[seq(1, nrow(wealth_data), 10), ], 
       aes(x = time, y = wealth, color = agent_type)) +
  geom_line(alpha = 0.3, size = 0.2) +
  stat_smooth(method = "loess", se = FALSE, size = 1) +
  scale_y_log10() +
  theme_minimal() +
  labs(title = "Agent Wealth Evolution by Type",
       x = "Time Period", y = "Wealth (log scale)",
       color = "Agent Type")

Time Series Visualizations

Market Regime Highlighting

# Create time series with regime highlighting
regime_colors <- c("1" = "lightgreen", "2" = "lightblue", "3" = "lightcoral")

# Plot spillover evolution with regimes
spillover_ts <- data.frame(
  time = 1:length(spillover_results$total_spillover),
  spillover = spillover_results$total_spillover,
  regime = sim_results$market_regimes[spillover_results$window_size:length(sim_results$market_regimes)]
)

ggplot(spillover_ts, aes(x = time, y = spillover)) +
  geom_line(size = 0.8) +
  geom_point(aes(color = as.factor(regime)), size = 1.5, alpha = 0.7) +
  scale_color_manual(values = regime_colors,
                     name = "Market Regime",
                     labels = c("Low Vol", "Normal", "High Vol")) +
  theme_minimal() +
  labs(title = "Total Spillover Evolution by Market Regime",
       x = "Time Window", y = "Total Spillover (%)")

Rolling Analysis Plots

# Create rolling correlation plot
rolling_data <- data.frame(
  time = 1:length(spillover_results$network_density),
  density = spillover_results$network_density,
  clustering = spillover_results$network_clustering
)

# Network metrics evolution
ggplot(rolling_data, aes(x = time)) +
  geom_line(aes(y = density, color = "Network Density"), size = 1) +
  geom_line(aes(y = clustering, color = "Clustering"), size = 1) +
  scale_color_manual(values = c("Network Density" = "blue", "Clustering" = "red")) +
  theme_minimal() +
  labs(title = "Network Structure Evolution",
       x = "Time Window", y = "Metric Value",
       color = "Metric")

Heatmap Visualizations

Spillover Heatmaps

# Enhanced spillover heatmap
spillover_heatmap <- plot_enhanced_spillover_heatmap(
  spillover_matrix = spillover_results$average_spillover_matrix,
  color_scheme = "RdYlBu",
  title = "Average Spillover Matrix"
)

print(spillover_heatmap)

Correlation Heatmaps

# Calculate correlation matrix
correlation_matrix <- cor(data, use = "complete.obs")

# Create correlation heatmap
library(reshape2)
correlation_long <- melt(correlation_matrix)

ggplot(correlation_long, aes(x = Var1, y = Var2, fill = value)) +
  geom_tile() +
  scale_fill_gradient2(low = "blue", mid = "white", high = "red",
                       midpoint = 0, limit = c(-1, 1),
                       name = "Correlation") +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  labs(title = "Asset Correlation Matrix",
       x = "Asset", y = "Asset")

Publication Dashboard

Multi-Panel Layout

# Create comprehensive dashboard
dashboard <- generate_publication_dashboard(
  simulation_results = sim_results,
  spillover_results = spillover_results,
  layout = "2x2"
)

print(dashboard)

Custom Dashboard Layout

# Create custom multi-panel plot
library(gridExtra)

# Panel 1: Network plot
p1 <- plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  layout = "stress",
  title = "A) Network Structure"
)

# Panel 2: Spillover evolution
p2 <- ggplot(spillover_ts, aes(x = time, y = spillover)) +
  geom_line() +
  theme_minimal() +
  labs(title = "B) Spillover Evolution", x = "Time", y = "Spillover (%)")

# Panel 3: Wealth inequality
wealth_gini <- data.frame(
  time = 1:length(sim_results$wealth_evolution_gini),
  gini = sim_results$wealth_evolution_gini
)

p3 <- ggplot(wealth_gini, aes(x = time, y = gini)) +
  geom_line() +
  theme_minimal() +
  labs(title = "C) Wealth Inequality", x = "Time", y = "Gini Coefficient")

# Panel 4: Agent performance
agent_performance <- sim_results$agent_performance_summary

p4 <- ggplot(agent_performance, aes(x = agent_type, y = mean_return, fill = agent_type)) +
  geom_col() +
  theme_minimal() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) +
  labs(title = "D) Agent Performance", x = "Agent Type", y = "Mean Return")

# Combine panels
grid.arrange(p1, p2, p3, p4, ncol = 2)

Interactive Visualizations

Plotly Integration

library(plotly)

# Convert ggplot to interactive plot
interactive_spillover <- ggplotly(
  ggplot(spillover_ts, aes(x = time, y = spillover, color = as.factor(regime))) +
    geom_line() +
    theme_minimal() +
    labs(title = "Interactive Spillover Evolution")
)

interactive_spillover

Dynamic Network Animation

# Create animated network (conceptual - requires additional packages)
# This would show network evolution over time
# animate_network_evolution(spillover_results, n_frames = 50)

Customization Options

Theme Customization

# Custom ggplot theme for publications
publication_theme <- theme_minimal() +
  theme(
    plot.title = element_text(size = 14, face = "bold"),
    axis.title = element_text(size = 12),
    axis.text = element_text(size = 10),
    legend.title = element_text(size = 11),
    legend.text = element_text(size = 10),
    panel.grid.minor = element_blank(),
    plot.background = element_rect(fill = "white", color = NA)
  )

# Apply to plots
plot_enhanced_network(
  spillover_results$average_spillover_matrix,
  layout = "stress"
) + publication_theme

Color Palette Customization

# Custom color palettes
custom_colors <- c("#1f77b4", "#ff7f0e", "#2ca02c", "#d62728", "#9467bd", "#8c564b")

# Apply to agent network
plot_agent_type_network(
  agents = agents,
  network_obj = network$networks$trading$graph,
  color_palette = custom_colors
)

Export and Saving

High-Quality Image Export

# Save high-resolution plots
ggsave(
  filename = "network_analysis.png",
  plot = network_plot,
  width = 10, height = 8,
  dpi = 300, bg = "white"
)

# Save PDF for publications
ggsave(
  filename = "network_analysis.pdf",
  plot = network_plot,
  width = 10, height = 8,
  device = "pdf"
)

Batch Export

# Export multiple visualizations
plots_list <- list(
  network = network_plot,
  spillover_evolution = ggplot(spillover_ts, aes(x = time, y = spillover)) + geom_line(),
  wealth_inequality = ggplot(wealth_gini, aes(x = time, y = gini)) + geom_line()
)

# Save all plots
for(name in names(plots_list)) {
  ggsave(
    filename = paste0(name, ".png"),
    plot = plots_list[[name]],
    width = 10, height = 6, dpi = 300
  )
}

Performance Tips

Large Network Visualization

# For large networks (>50 nodes):
# - Use simpler layouts (e.g., "circle")
# - Reduce edge transparency
# - Filter weak connections
# - Use node aggregation

# Filter weak spillovers
filtered_matrix <- spillover_results$average_spillover_matrix
filtered_matrix[filtered_matrix < quantile(filtered_matrix, 0.75)] <- 0

plot_enhanced_network(
  filtered_matrix,
  layout = "circle",
  edge_alpha = 0.6
)

Memory Optimization

# For large datasets:
# - Sample data points for time series
# - Use binning for scatter plots
# - Create plots in chunks
# - Clear plot objects after saving

# Sample for visualization
sample_indices <- seq(1, nrow(spillover_ts), by = 5)
spillover_sample <- spillover_ts[sample_indices, ]

# Create plot with sampled data
ggplot(spillover_sample, aes(x = time, y = spillover)) +
  geom_line()

Next Steps

Explore Network Analysis for spillover calculation details
Read Agent-Based Modeling for simulation components
Try Advanced Analysis for complex multi-asset workflows
Check Getting Started for basic usage examples

References

Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis.
Pedersen, T. L. (2021). ggraph: An Implementation of Grammar of Graphics for Graphs and Networks.
Garnier, S., et al. (2021). viridis: Colorblind-Friendly Color Maps for R.