# Load required libraries library(tidyverse) library(igraph) library(ggraph) library(colorspace) library(plotly) # Read the data data <- read_csv("Downloads/cgt_county_data_sep_29_2024.csv") # Filter data and calculate RCA data_filtered <- data %>% filter(!grepl("county not reported", county_name, ignore.case = TRUE)) # Create a matrix of industry co-occurrences industry_matrix <- data_filtered %>% filter(M == 1) %>% select(county_geoid, industry_code) %>% table() %>% as.matrix() # Calculate the co-occurrence matrix co_occurrence <- t(industry_matrix) %*% industry_matrix # Calculate the proximity matrix diag_values <- diag(co_occurrence) proximity <- co_occurrence / pmax(rep(diag_values, each = nrow(co_occurrence)), rep(diag_values, times = nrow(co_occurrence))) diag(proximity) <- 0 # Create an edge list from the proximity matrix edge_list <- which(proximity > quantile(proximity[proximity > 0], 0.95), arr.ind = TRUE) edge_df <- data.frame( from = rownames(proximity)[edge_list[,1]], to = colnames(proximity)[edge_list[,2]], weight = proximity[edge_list] ) # Create the graph object g <- graph_from_data_frame(edge_df, directed = FALSE) # Add node attributes V(g)$name <- data_filtered$industry_desc[match(V(g)$name, data_filtered$industry_code)] V(g)$sector <- substr(data_filtered$industry_code[match(V(g)$name, data_filtered$industry_desc)], 1, 2) # Create a color palette for sectors n_sectors <- length(unique(V(g)$sector)) sector_colors <- setNames( rainbow_hcl(n_sectors), sort(unique(V(g)$sector)) ) # Create the ggplot object with tooltips set.seed(42) # For reproducibility p <- ggraph(g, layout = "fr") + geom_edge_link(aes(edge_alpha = weight), show.legend = FALSE, edge_colour = "lightgray") + geom_node_point(aes(color = sector, text = name), size = 3) + # 'text' stores the industry name for tooltips geom_node_text(aes(label = name), repel = TRUE, size = 2, max.overlaps = 10) + scale_edge_alpha(range = c(0.1, 0.5)) + scale_color_manual(values = sector_colors) + theme_graph() + theme(legend.position = "none") + labs(title = "Industry Space: Co-location Based Proximity") # Convert ggplot to an interactive plotly object with hover info p_interactive <- ggplotly(p, tooltip = "text") # 'tooltip = "text"' ensures industry names appear on hover # Display the interactive plot p_interactive # Save the interactive plot as an HTML file htmlwidgets::saveWidget(p_interactive, "industry_space_interactive.html")

Including all NAICS six-digit industries.

September 19, 2024

By: Ben Feshbach | Date: September 19, 2024 | Data source: Clean Growth Tool county-level backend data