Chapter 7 Examples and Case Studies
This chapter provides practical examples and case studies demonstrating various uses of the ASA ABM v2.
7.1 Basic Examples
7.1.1 Example 1: Standard Simulation
A basic simulation with default parameters:
# Load the simulation engine
source("simulation/engine.R")
# Run a one-year simulation
results <- run_asa_simulation(
n_steps = 52, # Weekly steps
initial_size = 100, # Starting size
verbose = TRUE
)
# View final metrics
final_metrics <- tail(results$metrics, 1)
print(final_metrics)
# Plot organization size over time
library(ggplot2)
ggplot(results$metrics, aes(x = time, y = size)) +
geom_line(color = "darkblue", size = 1) +
labs(title = "Organization Size Over Time",
x = "Week", y = "Number of Employees") +
theme_minimal()7.1.2 Example 2: High-Turnover Environment
Simulating an organization with challenging conditions:
# Configure high-turnover parameters
high_turnover_params <- list(
turnover_threshold = -3, # Low satisfaction tolerance
growth_rate = 0.05, # Aggressive hiring
n_interactions_per_step = 2, # Limited interactions
interaction_window = 5 # Short memory
)
# Run simulation
volatile_results <- run_asa_simulation(
n_steps = 260,
initial_size = 50,
params = high_turnover_params
)
# Analyze turnover patterns
library(dplyr)
turnover_analysis <- volatile_results$metrics %>%
mutate(
period = ceiling(time / 13), # Quarterly
size_change = size - lag(size, default = 50)
) %>%
group_by(period) %>%
summarise(
avg_size = mean(size),
total_turnover = sum(size_change[size_change < 0]),
turnover_rate = abs(total_turnover) / avg_size
)
print(turnover_analysis)7.1.3 Example 3: Diversity-Focused Hiring
Testing different selection strategies:
# Compare selection strategies
strategies <- c("conscientiousness", "fit", "random")
strategy_results <- list()
for (strategy in strategies) {
set.seed(123) # For comparability
results <- run_asa_simulation(
n_steps = 156, # 3 years
initial_size = 100,
params = list(
selection_criteria = strategy,
growth_rate = 0.02
),
verbose = FALSE
)
strategy_results[[strategy]] <- results$metrics %>%
mutate(strategy = strategy)
}
# Combine and plot
library(tidyr)
combined_results <- bind_rows(strategy_results)
ggplot(combined_results, aes(x = time, y = identity_diversity,
color = strategy)) +
geom_line(size = 1) +
labs(title = "Identity Diversity by Selection Strategy",
x = "Time", y = "Shannon Diversity Index") +
theme_minimal() +
scale_color_brewer(palette = "Set1")7.2 Advanced Examples
7.2.1 Example 4: Parameter Sensitivity Analysis
Testing how different parameters affect outcomes:
# Define parameter grid
param_grid <- expand.grid(
growth_rate = c(0.01, 0.02, 0.05),
turnover_threshold = c(-10, -5, -2),
n_interactions = c(5, 10, 20)
)
# Run simulations
sensitivity_results <- list()
for (i in 1:nrow(param_grid)) {
params <- list(
growth_rate = param_grid$growth_rate[i],
turnover_threshold = param_grid$turnover_threshold[i],
n_interactions_per_step = param_grid$n_interactions[i]
)
results <- run_asa_simulation(
n_steps = 104, # 2 years
initial_size = 100,
params = params,
verbose = FALSE
)
# Extract key metrics
final_state <- tail(results$metrics, 1)
sensitivity_results[[i]] <- data.frame(
param_grid[i,],
final_size = final_state$size,
final_satisfaction = final_state$avg_satisfaction,
final_diversity = final_state$identity_diversity
)
}
# Analyze results
sensitivity_df <- bind_rows(sensitivity_results)
# Create heatmap
library(reshape2)
size_matrix <- dcast(sensitivity_df,
growth_rate ~ turnover_threshold,
value.var = "final_size",
fun.aggregate = mean)
# Plot heatmap
ggplot(melt(size_matrix), aes(x = factor(turnover_threshold),
y = factor(growth_rate),
fill = value)) +
geom_tile() +
geom_text(aes(label = round(value))) +
scale_fill_gradient2(low = "red", mid = "white", high = "green",
midpoint = 100) +
labs(title = "Final Organization Size by Parameters",
x = "Turnover Threshold", y = "Growth Rate") +
theme_minimal()7.2.2 Example 5: Intervention Analysis
Testing organizational interventions:
# Baseline simulation
baseline <- run_asa_simulation(
n_steps = 104,
initial_size = 100,
params = list(tag = "baseline"),
verbose = FALSE
)
# Intervention: Improve interaction quality
intervention <- run_asa_simulation(
n_steps = 104,
initial_size = 100,
params = list(
n_interactions_per_step = 15, # Triple interactions
interaction_window = 20, # Longer memory
tag = "intervention"
),
verbose = FALSE
)
# Compare results
comparison <- bind_rows(
baseline$metrics %>% mutate(scenario = "Baseline"),
intervention$metrics %>% mutate(scenario = "Intervention")
)
# Plot satisfaction trajectories
ggplot(comparison, aes(x = time, y = avg_satisfaction,
color = scenario)) +
geom_line(size = 1.2) +
labs(title = "Impact of Increased Interactions on Satisfaction",
x = "Time", y = "Average Satisfaction") +
theme_minimal() +
scale_color_manual(values = c("Baseline" = "gray50",
"Intervention" = "darkgreen"))7.3 Case Studies
7.3.1 Case Study 1: Startup Growth Dynamics
Modeling a rapidly growing startup:
# Startup parameters
startup_params <- list(
# Aggressive growth
growth_rate = 0.10,
hiring_frequency = 2,
n_new_applicants = 200,
# Culture-focused selection
selection_criteria = "fit",
# High interaction environment
n_interactions_per_step = 20,
interaction_window = 20,
# Moderate turnover tolerance
turnover_threshold = -5
)
# Run 2-year simulation
startup_sim <- run_asa_simulation(
n_steps = 104,
initial_size = 10, # Small founding team
params = startup_params
)
# Analyze growth phases
growth_analysis <- startup_sim$metrics %>%
mutate(
phase = case_when(
time <= 26 ~ "Founding",
time <= 52 ~ "Early Growth",
time <= 78 ~ "Scaling",
TRUE ~ "Maturing"
)
) %>%
group_by(phase) %>%
summarise(
start_size = first(size),
end_size = last(size),
growth_rate = (end_size - start_size) / start_size,
avg_satisfaction = mean(avg_satisfaction),
avg_diversity = mean(identity_diversity)
)
print(growth_analysis)
# Visualize growth trajectory
ggplot(startup_sim$metrics, aes(x = time)) +
geom_line(aes(y = size), color = "blue", size = 1) +
geom_line(aes(y = avg_satisfaction * 100),
color = "green", size = 1, linetype = "dashed") +
scale_y_continuous(
name = "Organization Size",
sec.axis = sec_axis(~./100, name = "Avg Satisfaction")
) +
labs(title = "Startup Growth and Satisfaction",
x = "Week") +
theme_minimal()7.3.2 Case Study 2: Merger Integration
Simulating the integration of two organizational cultures:
# Create two distinct organizations
org_a <- create_organization(
n_agents = 80,
identity_categories = c("A", "A", "A", "B", "B") # A-dominant
)
org_b <- create_organization(
n_agents = 60,
identity_categories = c("C", "C", "D", "D", "E") # Different culture
)
# Merge organizations
merged_org <- rbind(org_a, org_b)
merged_org$agent_id <- paste0("merged_", seq_len(nrow(merged_org)))
# Run post-merger simulation
merger_params <- list(
growth_rate = 0, # No hiring during integration
n_interactions_per_step = 15, # Encourage mixing
turnover_threshold = -8 # Some tolerance for dissatisfaction
)
# Initialize properly
source("simulation/engine.R")
merger_sim <- list(
final_organization = merged_org,
metrics = data.table(),
parameters = merger_params
)
# Track cultural integration
integration_metrics <- data.frame()
for (month in 1:12) {
# Run one month
month_results <- run_asa_simulation(
n_steps = 4,
initial_size = nrow(merged_org),
params = merger_params,
verbose = FALSE
)
# Calculate integration index
interactions <- initialize_interactions(month_results$final_organization)
# Store metrics
integration_metrics <- rbind(
integration_metrics,
data.frame(
month = month,
size = get_organization_size(month_results$final_organization),
diversity = calculate_identity_diversity(month_results$final_organization),
satisfaction = calculate_average_satisfaction(month_results$final_organization)
)
)
# Update for next iteration
merged_org <- month_results$final_organization
}
# Plot integration progress
ggplot(integration_metrics, aes(x = month)) +
geom_line(aes(y = satisfaction), color = "blue", size = 1) +
geom_line(aes(y = diversity), color = "green", size = 1) +
geom_point(aes(y = satisfaction), color = "blue") +
geom_point(aes(y = diversity), color = "green") +
labs(title = "Post-Merger Integration Metrics",
x = "Months Post-Merger",
y = "Metric Value") +
theme_minimal()7.3.3 Case Study 3: Remote Work Transition
Modeling the impact of remote work on organizational dynamics:
# Pre-remote baseline
pre_remote <- run_asa_simulation(
n_steps = 52,
initial_size = 200,
params = list(
n_interactions_per_step = 10,
interaction_window = 10
),
verbose = FALSE
)
# Remote work scenario (reduced interactions)
remote_params <- list(
n_interactions_per_step = 3, # Fewer spontaneous interactions
interaction_window = 15, # But longer-lasting connections
turnover_threshold = -7, # Slightly more tolerance
growth_rate = 0.005 # Slower growth
)
remote_sim <- run_asa_simulation(
n_steps = 52,
initial_size = 200,
params = remote_params,
verbose = FALSE
)
# Hybrid scenario
hybrid_params <- list(
n_interactions_per_step = 6, # Moderate interactions
interaction_window = 12,
turnover_threshold = -8,
growth_rate = 0.015
)
hybrid_sim <- run_asa_simulation(
n_steps = 52,
initial_size = 200,
params = hybrid_params,
verbose = FALSE
)
# Compare scenarios
scenarios <- bind_rows(
pre_remote$metrics %>% mutate(scenario = "Office"),
remote_sim$metrics %>% mutate(scenario = "Remote"),
hybrid_sim$metrics %>% mutate(scenario = "Hybrid")
)
# Multi-panel comparison
library(gridExtra)
p1 <- ggplot(scenarios, aes(x = time, y = avg_satisfaction,
color = scenario)) +
geom_line(size = 1) +
labs(title = "Satisfaction", y = "Average") +
theme_minimal() +
theme(legend.position = "bottom")
p2 <- ggplot(scenarios, aes(x = time, y = identity_diversity,
color = scenario)) +
geom_line(size = 1) +
labs(title = "Diversity", y = "Shannon Index") +
theme_minimal() +
theme(legend.position = "bottom")
p3 <- ggplot(scenarios, aes(x = time, y = size,
color = scenario)) +
geom_line(size = 1) +
labs(title = "Organization Size", y = "Employees") +
theme_minimal() +
theme(legend.position = "bottom")
grid.arrange(p1, p2, p3, ncol = 3)7.4 Best Practices from Examples
7.4.1 1. Parameter Selection
- Start with default values and adjust gradually
- Consider realistic ranges for your context
- Document parameter choices and rationale
7.4.2 2. Analysis Approach
- Always visualize time series data
- Compare multiple scenarios
- Calculate summary statistics for key periods
- Consider both individual and aggregate metrics
7.5 Code Snippets for Common Tasks
7.5.2 Custom Metrics
# Add custom metric calculation
calculate_custom_metric <- function(org) {
# Example: Personality homogeneity
personality_vars <- c("openness", "conscientiousness",
"extraversion", "agreeableness",
"emotional_stability")
homogeneity <- org[is_active == TRUE,
lapply(.SD, function(x) 1/sd(x)),
.SDcols = personality_vars]
return(mean(as.numeric(homogeneity)))
}7.5.3 Visualization Helpers
# Create standard plot theme
theme_asa <- function() {
theme_minimal() +
theme(
plot.title = element_text(face = "bold"),
legend.position = "bottom",
panel.grid.minor = element_blank()
)
}
# Time series with confidence bands
plot_with_ci <- function(results_list, metric) {
# Calculate mean and CI across replications
combined <- bind_rows(lapply(seq_along(results_list),
function(i) {
results_list[[i]]$metrics %>%
mutate(rep = i)
}))
summary_stats <- combined %>%
group_by(time) %>%
summarise(
mean_val = mean(get(metric)),
lower_ci = quantile(get(metric), 0.025),
upper_ci = quantile(get(metric), 0.975)
)
ggplot(summary_stats, aes(x = time)) +
geom_ribbon(aes(ymin = lower_ci, ymax = upper_ci),
alpha = 0.3) +
geom_line(aes(y = mean_val), size = 1) +
labs(y = metric) +
theme_asa()
}