library(ggplot2) ggplot(merged_data, aes(x = Decision_Class, y = AI_Capability_Index, fill = Decision_Class)) + geom_boxplot() + theme_minimal() + labs(title = "AI Capability Index by Decision Class", x = "Decision Class", y = "AI Capability Index") # Visualize Public Sector Innovation across decision classes merged_data_long <- pivot_longer(merged_data, cols = c(AI_Capability_Index, Public_Sector_Innovation, Government.Strategy), names_to = "Metric", values_to = "Value") # Combined boxplot comparing AI Capability Index, Public Sector Innovation, and Government.Strategy ggplot(merged_data_long, aes(x = Decision_Class, y = Value, fill = Metric)) + geom_boxplot(position = position_dodge(0.8)) + theme_minimal() + labs(title = "Comparison of AI Capability, Public Sector Innovation, and Government Strategy by Decision Class", x = "Decision Class", y = "Development Factor", fill = "Metric") + theme(legend.title = element_text(face = "bold"), plot.title = element_text(hjust = 0.5))

install.packages("rworldmap") library(rworldmap) # Merge the classification data with a world map map_data <- joinCountryData2Map(merged_data, joinCode = "NAME", nameJoinColumn = "Country") # Visualize the classification on a world map mapCountryData(map_data, nameColumnToPlot = "Economic_Development", mapTitle = "Developed vs Developing Countries", catMethod = "categorical", colourPalette = c("Developed" = "#2E86C1", "Developing" = "#F39C12"))

plot(merged_data$norm_talent, merged_data$AI_Capability_Index, xlab = "Normalized Talent", ylab = "AI Capability Index")

boxplot(norm_talent ~ Country, data = merged_data, main = "Normalized Talent by Country", xlab = "Country", ylab = "Normalized Talent")

summary(merged_data$Decision) hist(merged_data$Decision, main = "Distribution of Decision Variable", xlab = "Decision Score")

library(lime) explainer <- lime(train_data_2, logistic_model) explanation <- explain(test_data_2[sample(1:nrow(test_data_2), 10), ], explainer, n_labels = 1, n_features = 2) plot_features(explanation)

library(lime) explainer <- lime(train_data_1[, c("AI_Capability_Index", "Public_Sector_Innovation")], model_politic) # Explain a few instances from the test set explanation <- explain(test_data_1[1:5, c("AI_Capability_Index", "Public_Sector_Innovation")], explainer, n_labels = 1, n_features = 3) # Visualize the LIME explanations plot_features(explanation)

###classficication # Load necessary libraries library(randomForest) library(caret) # Define the predictors and target variable predictors <- merged_data[, c("Talent", "Infrastructure", "Operating.Environment", "Research", "Development", "Government.Strategy", "AI_Capability_Index", "Public_Sector_Innovation")] target <- merged_data$Political.regime # Train-test split set.seed(123) trainIndex <- createDataPartition(target, p = .8, list = FALSE, times = 1) trainData <- merged_data[trainIndex, ] testData <- merged_data[-trainIndex, ] # Fit a random forest model rf_model <- randomForest(as.factor(Political.regime) ~ Talent + Infrastructure + Operating.Environment + Research + Development + Government.Strategy + AI_Capability_Index + Public_Sector_Innovation, data = trainData, ntree = 500, mtry = 3, importance = TRUE) # Model Evaluation predictions <- predict(rf_model, testData) confusionMatrix(predictions, testData$Political.regime) # View feature importance importance(rf_model) varImpPlot(rf_model)

# Histograms for AI_Capability_Index and Public_Sector_Innovation hist(merged_data$AI_Capability_Index, main = "AI Capability Index Distribution", xlab = "AI Capability Index", col = "lightblue")

hist(merged_data$Public_Sector_Innovation, main = "Public Sector Innovation Distribution", xlab = "Public Sector Innovation", col = "lightgreen")

ggplot(merged_data, aes(x = reorder(Country, -AI_Capability_Index), y = AI_Capability_Index, fill = as.factor(Decision))) + geom_bar(stat = "identity") + coord_flip() + # Flip for horizontal bars labs(title = "Countries Ranked by AI Capability and Geopolitical Decision Influence", x = "Country", y = "AI Capability Index", fill = "Decision Impact") + theme_minimal()

> data_to_cluster <- merged_data[, c("AI_Capability_Index", "Talent", "Public_Sector_Innovation")] > > data_scaled <- scale(data_to_cluster) > pheatmap(data_scaled, + cluster_rows = TRUE, cluster_cols = TRUE, + main = "Clustered Heatmap of AI Capability and Geopolitical Factors", + display_numbers = TRUE)

decision_tree <- rpart(Political.regime ~ Talent + Infrastructure + AI_Capability_Index + Public_Sector_Innovation, data = df, method = "class") # Plot the decision tree rpart.plot(decision_tree, type = 2, extra = 104, under = TRUE, faclen = 0, cex = 0.8)

# Add custom annotations to describe the clusters ggplot(hclust.project2D, aes(x = PC1, y = PC2)) + geom_point(aes(shape = cluster_label, color = cluster_label), size = 3) + geom_text(aes(label = country, color = cluster_label), hjust = 0, vjust = 1, size = 3) + geom_polygon(data = hclust.hull, aes(group = cluster, fill = as.factor(cluster)), alpha = 0.4, linetype = 0) + theme_minimal() + labs(title = "Geopolitical Decisions Based on AI Capabilities", x = "Principal Component 1 (AI Capability)", y = "Principal Component 2 (Government Strategy)", color = "Cluster", shape = "Cluster") + scale_fill_manual(values = c("#FF9999", "#99CCFF", "#66CC66"), labels = c("AI Leaders", "Emerging AI Players", "Tech Adopters")) + theme(text = element_text(size = 16)) + annotate("text", x = 5, y = -2, label = "AI Leaders: Advanced AI capabilities, \nForm AI alliances, Lead AI standards", color = "#FF9999", size = 5, hjust = 0) + annotate("text", x = -4, y = 3, label = "Emerging AI Players: Focus on AI regulation, \nBuild partnerships for growth", color = "#99CCFF", size = 5, hjust = 0) + annotate("text", x = -6, y = -5, label = "Tech Adopters: Focus on cybersecurity, \nEnsure digital sovereignty", color = "#66CC66", size = 5, hjust = 0)

library(ggplot2) # Plot the clusters with meaningful labels ggplot(hclust.project2D, aes(x = PC1, y = PC2)) + geom_point(aes(shape = cluster_label, color = cluster_label), size = 3) + geom_text(aes(label = country, color = cluster_label), hjust = 0, vjust = 1, size = 3) + geom_polygon(data = hclust.hull, aes(group = cluster, fill = as.factor(cluster)), alpha = 0.4, linetype = 0) + theme_minimal() + labs(title = "Geopolitical Decisions Based on AI Capabilities", x = "Principal Component 1 (AI Capability)", y = "Principal Component 2 (Government Strategy)", color = "Cluster", shape = "Cluster") + scale_fill_manual(values = c("#FF9999", "#99CCFF", "#66CC66"), labels = c("AI Leaders", "Emerging AI Players", "Tech Adopters")) + theme(text = element_text(size = 16))

# data prepare for clustering visualization princ <- prcomp(scaled_data) nComp <- 2 project2D <- as.data.frame(predict(princ, newdata=scaled_data)[,1:nComp]) hclust.project2D <- cbind(project2D, cluster=as.factor(groups), country=df$Country) head(hclust.project2D) library('grDevices') find_convex_hull <-function(proj2Ddf,groups) { do.call(rbind, lapply(unique(groups), FUN= function(c){ f<-subset(proj2Ddf,cluster==c); f[chull(f),] } ) ) } hclust.hull <-find_convex_hull(hclust.project2D,groups) library(ggplot2) ggplot(hclust.project2D, aes(x=PC1,y=PC2)) + geom_point(aes(shape=cluster,color=cluster)) + geom_text(aes(label=country,color=cluster),hjust=0,vjust=1, size=3) + geom_polygon(data=hclust.hull, aes(group=cluster,fill=as.factor(cluster)), alpha=0.4, linetype=0) + theme(text=element_text(size=20))

# Add cluster information to the dataset df$Cluster <- groups # Plot the boxplot ggplot(df, aes(x = as.factor(Cluster), y = AI_Capability_Index, fill = as.factor(Cluster))) + geom_boxplot() + labs(title = "Distribution of AI Capability Index by Cluster", x = "Cluster", y = "AI Capability Index") + theme_minimal()

# Calculate the mean AI_Capability_Index for each cluster and region library(dplyr) cluster_summary <- df %>% mutate(cluster = groups) %>% group_by(cluster, Region) %>% summarise(mean_AICapability = mean(AI_Capability_Index, na.rm = TRUE)) # Plot the bar plot library(ggplot2) ggplot(cluster_summary, aes(x = Region, y = mean_AICapability, fill = as.factor(cluster))) + geom_bar(stat = "identity", position = "dodge") + labs(title = "AI Capability Index by Region and Cluster", x = "Region", y = "Mean AI Capability Index", fill = "Cluster") + theme_minimal()

Dendrogram is a tree-like diagram that shows the hierarchical structure of the clusters library(dynamicTreeCut) dist_matrix <- dist(merged_data[, -1]) # Assuming the first column is 'Region' hclust_model <- hclust(dist_matrix, method = "ward.D2") options(repr.plot.width = 16, repr.plot.height = 10) plot(hclust_model, labels = merged_data$Region, main = "Hierarchical Clustering Dendrogram", xlab = "Region", ylab = "Height", sub = "", cex = 0.4, las = 2) rect.hclust(hclust_model, k = 5, border = "red") dendro_colors <- cutreeDynamic(hclust_model, distM = as.matrix(dist_matrix), method = "tree") plot(hclust_model, labels = merged_data$Region, main = "Hierarchical Clustering Dendrogram with Dynamic Cut", xlab = "Region", ylab = "Height", sub = "", cex = 0.4, las = 2) rect.hclust(hclust_model, k = 5, border = "red")

> library(rpart) > library(rpart.plot) > > # Train decision tree model > decision_tree <- rpart(Decision ~ AI_Capability_Index + Public_Sector_Innovation + Talent + Infrastructure + Government.Strategy, + data = merged_data, method = "class") > > # Plot the decision tree > rpart.plot(decision_tree, type = 3, extra = 104, fallen.leaves = TRUE, main = "Geopolitical Decisions")

ggplot(merged_data, aes(x = Political.regime, y = AI_Capability_Index)) + geom_boxplot() + theme_minimal() Compare the distribution of AI_Capability_Index across different Political.regime categories.

library(ggplot2) ggplot(merged_data, aes(x = Infrastructure, y = AI_Capability_Index)) + geom_point() + geom_smooth(method = "lm") + theme_minimal()

AI capability vs Total.score have strong relationships ggplot(merged_data, aes(x = `Total score`, y = AI_Capability_Index)) + geom_point(color = "blue", alpha = 0.7) + geom_smooth(method = "lm", color = "red", se = FALSE) + labs(title = "Relationship Between AI Capability Index and Total Score", x = "Total Score", y = "AI Capability Index") + theme_minimal()

see how the AI capability index varies across different income groups

analyze how AI capability varies by region or country cluster

plot shows correlations between Government AI Strategy, AI capability index, and Political Regime (numerically encoded). This will help identify linear relationships between these factors

plot shows the Government AI Strategy, AI Readiness, and Political Regime in a 3D space, allowing you to see how these factors interact. You can rotate and explore different angles of the 3D scatter plot

show the mean AI readiness for each combination of Political Regime and Government AI Strategy. You can spot which combinations are most effective at driving AI readiness.

This visualization will show pairwise scatter plots for each combination of the variables: Government Strategy vs AI Readiness. Government Strategy vs Political Regime. AI Readiness vs Political Regime. The plot will help you identify relationships, trends, and interactions between the three variables. The color coding by Political Regime helps distinguish different regimes visually

visualization helps you understand the interaction between Government AI Strategy and Political Regime, showing how different regimes approach AI strategy and how it impacts AI readiness

plot shows the full distribution of AI readiness for each political regime, highlighting areas of high density, median values, and variability.

heatmap shows the strength of the correlations between political regimes, AI readiness, government AI strategy, and income group. This helps to identify patterns (e.g., whether certain regimes are highly correlated with high AI readiness) ggplot(melted_corr, aes(Var1, Var2, fill = value)) + geom_tile() + scale_fill_gradient2(low = "blue", high = "red", mid = "white", midpoint = 0) + labs(title = "Correlation Heatmap of AI Readiness and Political Regime") + theme_minimal()

plot will show the distribution of AI readiness for each political regime, helping to compare how AI readiness differs between regimes (e.g., do liberal democracies have higher AI readiness than closed autocracies?)

Decision trees classify data by creating a tree-like model of decisions. The rpart package can be used to create a decision tree.

K-means algorithm follows an iterative process to assign data points to K clusters, where K is the number of clusters specified by the user

Linear Regression (e.g., Impact of Government Strategy on AI Metrics) Linear regression helps quantify the relationship between geopolitical factors (e.g., government strategy) and AI metrics (e.g., infrastructure, research).

You can also visualize trends across different categories using faceting and grouping in ggplot2.

Interactive Visualization with ggplot2 Use ggplot2 for more flexible and polished visualizations.

If your data contains time information (e.g., years of projects in the public sector data), you can visualize trends over time.

For numeric variables, a pair plot helps you visualize relationships between all variables simultaneously. pairs(~ Talent + Infrastructure + Development + `Total score`, data=merged_data, main="Pairwise Scatter Plot")

# Visualize the correlation matrix library(corrplot) corrplot(cor_matrix, method="color", type="upper", tl.col="black", tl.srt=45)

mosaicplot(table(ai_data$Political.regime, ai_data$Region), main="Region vs Political Regime", col=c("skyblue", "pink", "yellow"))

plot(merged_data$Development, merged_data$AI_Index_Score, main="AI Index score vs Development", xlab="Development", ylab="Total Score", pch=19, col="blue")

Distribution of Income Groups