library(ggplot2)
library(stringr)
library(tibble)
# --- Register Times New Roman for Windows graphics ---
windowsFonts(`Times New Roman` = windowsFont("Times New Roman"))
setwd("C:/Users/hbx29/OneDrive - Texas State University/Das, Subasish's files - FARSPaper/ARM_LIC/FARS_4362_papers/Ramp/LandUse/Data")
# -----------------------------
# 1) Load ALL rules dataset (cloud)
# -----------------------------
file_path <- "LIC S C lift_urban.xlsx"
df <- read_excel(file_path, sheet = "Sheet2")
# Fix empty/odd column names
names(df) <- trimws(names(df))
names(df)[names(df) == "" | is.na(names(df))] <- "blank_col"
names(df) <- make.names(names(df), unique = TRUE)
# Auto-rename common variations
rename_map <- list(
support = c("support", "Support", "S"),
confidence = c("confidence", "Confidence", "C"),
lift = c("lift", "Lift", "L"),
LIC = c("LIC", "Lic", "lic")
)
for (newname in names(rename_map)) {
hit <- intersect(rename_map[[newname]], names(df))
if (!(newname %in% names(df)) && length(hit) > 0) {
df <- df %>% rename(!!newname := all_of(hit[1]))
}
}
# ---- Thresholds ----
SUP_MIN <- 0.10
CONF_MIN <- 0.35 # for rural it will be .30
LIFT_MIN <- 1.00
LIC_MIN <- 1.00
df <- df %>%
mutate(
support = as.numeric(support),
confidence = as.numeric(confidence),
lift = as.numeric(lift),
LIC = as.numeric(LIC)
) %>%
filter(!is.na(support), !is.na(confidence), !is.na(lift), !is.na(LIC)) %>%
mutate(LIC_group = ifelse(LIC >= LIC_MIN, "LIC \u2265 1.0", "LIC < 1.0"))
# -----------------------------
# 2) Manual highlighted rules (from your table/image) [UPDATED: M1–M5 and U1–U10]
# -----------------------------
manual_rules <- tribble(
~RuleID, ~S, ~C, ~L, ~LIC,
"M1", 0.193, 0.484, 1.418, NA,
"U1", 0.153, 0.520, 1.523, 1.074,
"U2", 0.100, 0.535, 1.565, 1.104,
"M2", 0.262, 0.385, 1.127, NA,
"U3", 0.153, 0.520, 1.523, 1.351,
"U4", 0.107, 0.428, 1.254, 1.113,
"U5", 0.100, 0.535, 1.565, 1.389,
"M3", 0.235, 0.632, 1.088, NA,
"U6", 0.107, 0.693, 1.195, 1.098,
"U7", 0.106, 0.642, 1.106, 1.017,
"M4", 0.166, 0.678, 1.079, NA,
"U8", 0.103, 0.723, 1.151, 1.067,
"U9", 0.123, 0.686, 1.092, 1.012,
"M5", 0.188, 0.750, 1.102, NA,
"U10", 0.105, 0.754, 1.109, 1.006
) %>%
mutate(
support = S, confidence = C, lift = L,
# Force M rules into LIC ≥ 1 panel (since LIC missing)
LIC_for_facet = ifelse(str_detect(RuleID, "^M"), 1.0001, LIC),
LIC_group = ifelse(LIC_for_facet >= LIC_MIN, "LIC \u2265 1.0", "LIC < 1.0"),
rule_type = ifelse(str_detect(RuleID, "^M"), "Base Rules: M1\u2013M5", "Associate Rules: U1\u2013U10")
)
# -----------------------------
# 3) Facet titles with counts from FULL dataset
# -----------------------------
facet_counts <- df %>%
group_by(LIC_group) %>%
summarise(rules = n(), .groups = "drop") %>%
mutate(facet_lab = paste0(LIC_group, ", Rules = ", rules))
lab_map <- setNames(facet_counts$facet_lab, facet_counts$LIC_group)
# -----------------------------
# 4) Plot (no point labels)
# -----------------------------
p <- ggplot() +
geom_point(
data = df,
aes(x = support, y = lift, size = confidence, color = confidence),
alpha = 0.80
) +
geom_vline(xintercept = SUP_MIN, linetype = "dashed", linewidth = 1.2) +
geom_hline(yintercept = LIFT_MIN, linetype = "solid", linewidth = 1.2) +
geom_point(
data = manual_rules,
aes(x = support, y = lift, shape = rule_type, fill = rule_type),
size = 7, color = "black", stroke = 1.6, alpha = 1
) +
facet_wrap(~ LIC_group, nrow = 1, labeller = as_labeller(lab_map)) +
scale_color_gradient(low = "grey85", high = "red3", name = "confidence") +
scale_size_continuous(range = c(2.5, 3), name = "confidence\n(bubble size)") +
scale_fill_manual(
name = "",
values = c("Base Rules: M1\u2013M5" = "dodgerblue3",
"Associate Rules: U1\u2013U10" = "goldenrod2"),
breaks = c("Base Rules: M1\u2013M5", "Associate Rules: U1\u2013U10")
) +
scale_shape_manual(
name = "",
values = c("Base Rules: M1\u2013M5" = 22,
"Associate Rules: U1\u2013U10" = 21),
breaks = c("Base Rules: M1\u2013M5", "Associate Rules: U1\u2013U10")
) +
labs(x = "support", y = "lift") +
theme_bw(base_size = 30) +
theme(
text = element_text(family = "Times New Roman"),
legend.position = "right",
strip.text = element_text(face = "bold", size = 30),
axis.title = element_text(size = 30),
axis.text = element_text(size = 28),
legend.text = element_text(size = 26),
panel.grid.minor = element_blank()
) +
guides(
fill = guide_legend(order = 1, override.aes = list(size = 7, color = "black")),
shape = guide_legend(order = 1, override.aes = list(size = 7, color = "black")),
color = guide_colorbar(order = 2),
size = guide_legend(order = 3)
)
print(p)
