Performance Visualization
Now let’s create comprehensive visualizations to compare the strategies.
Overall Performance Comparison
# Side-by-side performance overview
overview_A <- analyzer_A$visualize_performance("overview") +
ggtitle("Strategy A: Broad COVID Research") +
theme(plot.title = element_text(size = 12)) +
ylim(0, 1)
overview_B <- analyzer_B$visualize_performance("overview") +
ggtitle("Strategy B: Targeted Clinical COVID") +
theme(plot.title = element_text(size = 12)) +
ylim(0, 1)
if (requireNamespace("patchwork", quietly = TRUE)) {
combined_overview <- overview_A + overview_B +
plot_annotation(title = "COVID Search Strategy Performance Comparison",
subtitle = "Broad vs. Targeted Clinical Approaches")
print(combined_overview)
} else {
print(overview_A)
print(overview_B)
}
Direct Metric Comparison
# Create direct comparison plot
comparison_data <- data.frame(
Strategy = c("Broad COVID Research", "Targeted Clinical COVID"),
Precision = c(metrics_A$precision_recall$precision, metrics_B$precision_recall$precision),
Recall = c(metrics_A$precision_recall$recall, metrics_B$precision_recall$recall),
F1_Score = c(metrics_A$precision_recall$f1_score, metrics_B$precision_recall$f1_score),
stringsAsFactors = FALSE
)
# Reshape for plotting
comparison_long <- rbind(
data.frame(Strategy = comparison_data$Strategy, Metric = "Precision", Value = comparison_data$Precision),
data.frame(Strategy = comparison_data$Strategy, Metric = "Recall", Value = comparison_data$Recall),
data.frame(Strategy = comparison_data$Strategy, Metric = "F1_Score", Value = comparison_data$F1_Score)
)
comparison_plot <- ggplot(comparison_long, aes(x = Metric, y = Value, fill = Strategy)) +
geom_col(position = "dodge", alpha = 0.8, width = 0.7) +
geom_text(aes(label = round(Value, 3)), position = position_dodge(width = 0.7), vjust = -0.5) +
scale_fill_manual(values = c("Broad COVID Research" = "#2E86AB", "Targeted Clinical COVID" = "#A23B72")) +
labs(title = "Direct Performance Metric Comparison",
subtitle = "Trade-offs: Broad strategy shows higher recall, Targeted shows higher precision",
y = "Score", x = "Performance Metric") +
theme_minimal() +
theme(legend.position = "bottom") +
ylim(0, 1)
print(comparison_plot)
Overlap Analysis Visualization
# Create overlap visualization
overlap_data <- data.frame(
Category = c("Broad Strategy Only", "Overlap", "Targeted Strategy Only"),
Count = c(strategy_comparison$overlap_analysis$unique_to_strategy1,
strategy_comparison$overlap_analysis$overlap_count,
strategy_comparison$overlap_analysis$unique_to_strategy2),
stringsAsFactors = FALSE
)
overlap_data$Percentage <- overlap_data$Count / strategy_comparison$overlap_analysis$total_unique * 100
overlap_plot <- ggplot(overlap_data, aes(x = Category, y = Count, fill = Category)) +
geom_col(alpha = 0.8, width = 0.7) +
geom_text(aes(label = paste0(Count, "\n(", round(Percentage, 1), "%)")), vjust = 0.5, size = 3.5) +
scale_fill_manual(values = c("Broad Strategy Only" = "#2E86AB",
"Overlap" = "#F18F01",
"Targeted Strategy Only" = "#A23B72")) +
labs(title = "Article Retrieval Overlap Analysis",
subtitle = "Complementary nature of broad vs. targeted approaches",
y = "Number of Articles", x = "Category") +
theme_minimal() +
theme(legend.position = "none",
axis.text.x = element_text(angle = 45, hjust = 1))
print(overlap_plot)
Temporal Distribution Comparison
# Temporal comparison
temporal_A_plot <- analyzer_A$visualize_performance("temporal") +
ggtitle("Strategy A: Temporal Distribution") +
theme(plot.title = element_text(size = 12))
temporal_B_plot <- analyzer_B$visualize_performance("temporal") +
ggtitle("Strategy B: Temporal Distribution") +
theme(plot.title = element_text(size = 12))
if (requireNamespace("patchwork", quietly = TRUE)) {
combined_temporal <- temporal_A_plot + temporal_B_plot +
plot_annotation(title = "Temporal Distribution Comparison")
print(combined_temporal)
} else {
print(temporal_A_plot)
print(temporal_B_plot)
}
