# ============================================================
# Chapter 11 Graphs (R / RStudio) — dataset_study.csv
# X = study_hours, Y = grade
# Produces ALL recommended graphs + saves PNGs for Word
# ============================================================

rm(list = ls())
options(scipen = 999)

# Packages
library(ggplot2)

# 1) Load data (adjust path if needed)
df <- read.csv("C:/Users/Klajd Koskija/Downloads/dataset_study.csv")
df$study_hours <- as.numeric(df$study_hours)
df$grade      <- as.numeric(df$grade)
df <- na.omit(df)

# Quick check
stopifnot(all(c("study_hours","grade") %in% names(df)))
n <- nrow(df)
cat("Rows:", n, "\n")

# 2) Fit the model
fit <- lm(grade ~ study_hours, data = df)
df$yhat <- fitted(fit)
df$res  <- resid(fit)

# Useful summary numbers
X <- df$study_hours
Y <- df$grade
Xbar <- mean(X)
Sxx  <- sum((X - Xbar)^2)
SSE  <- sum(df$res^2)
df_res <- n - 2
MSE <- SSE / df_res
tcrit <- qt(0.975, df = df_res)

cat("\nModel:\n")
print(summary(fit))

# 3) Create output folder for Word figures
out_dir <- "chapter11_figures"
if (!dir.exists(out_dir)) dir.create(out_dir)

# Helper to save ggplot nicely
save_plot <- function(p, filename, w=7.5, h=5.2, dpi=300){
  ggsave(file.path(out_dir, filename), plot = p, width = w, height = h, dpi = dpi)
}

# ============================================================
# FIGURE 1: Scatterplot (foundation)
# ============================================================
p1 <- ggplot(df, aes(x = study_hours, y = grade)) +
  geom_point(alpha = 0.25) +
  labs(
    title = "Figure 1. Scatterplot: Grade vs Study Hours",
    x = "Study Hours (X)",
    y = "Grade (Y)"
  )
print(p1)
save_plot(p1, "Fig1_Scatterplot.png")

# ============================================================
# FIGURE 2: Scatterplot + Regression Line + 95% CI band
# ============================================================
p2 <- ggplot(df, aes(x = study_hours, y = grade)) +
  geom_point(alpha = 0.20) +
  geom_smooth(method = "lm", se = TRUE) +
  labs(
    title = "Figure 2. Regression Line with 95% Confidence Band",
    x = "Study Hours (X)",
    y = "Grade (Y)"
  )
print(p2)
save_plot(p2, "Fig2_RegressionLine_CI.png")

# ============================================================
# FIGURE 3: Residuals vs Fitted (assumptions check)
# ============================================================
p3 <- ggplot(df, aes(x = yhat, y = res)) +
  geom_point(alpha = 0.25) +
  geom_hline(yintercept = 0, linewidth = 0.7) +
  labs(
    title = "Figure 3. Residuals vs Fitted Values",
    x = "Fitted Values (Ŷ)",
    y = "Residuals (e = Y - Ŷ)"
  )
print(p3)
save_plot(p3, "Fig3_Residuals_vs_Fitted.png")

# ============================================================
# FIGURE 4: Histogram of Residuals (normality check)
# ============================================================
p4 <- ggplot(df, aes(x = res)) +
  geom_histogram(bins = 40) +
  labs(
    title = "Figure 4. Histogram of Residuals",
    x = "Residuals",
    y = "Count"
  )
print(p4)
save_plot(p4, "Fig4_Residuals_Hist.png")

# ============================================================
# FIGURE 5: Normal Q–Q Plot of Residuals (stronger normality check)
# (Base R plot saved as PNG)
# ============================================================
png(file.path(out_dir, "Fig5_QQplot_Residuals.png"), width = 1100, height = 800, res = 150)
qqnorm(df$res, main = "Figure 5. Normal Q–Q Plot of Residuals")
qqline(df$res, lwd = 2)
dev.off()

# Also display in RStudio
qqnorm(df$res, main = "Figure 5. Normal Q–Q Plot of Residuals")
qqline(df$res, lwd = 2)

# ============================================================
# FIGURE 6: Regression line + 95% Mean CI band + 95% Prediction band
# (This is the “bonus” plot lecturers LOVE)
# ============================================================

# Build a grid of x values
x_grid <- seq(min(df$study_hours), max(df$study_hours), length.out = 300)
newd <- data.frame(study_hours = x_grid)

# Predict mean response CI and prediction interval
pred_mean <- predict(fit, newdata = newd, interval = "confidence", level = 0.95)
pred_ind  <- predict(fit, newdata = newd, interval = "prediction",  level = 0.95)

bands <- data.frame(
  study_hours = x_grid,
  fit = pred_mean[,"fit"],
  mean_lwr = pred_mean[,"lwr"],
  mean_upr = pred_mean[,"upr"],
  pred_lwr = pred_ind[,"lwr"],
  pred_upr = pred_ind[,"upr"]
)

p6 <- ggplot() +
  geom_point(data = df, aes(x = study_hours, y = grade), alpha = 0.15) +
  geom_line(data = bands, aes(x = study_hours, y = fit), linewidth = 1) +
  geom_line(data = bands, aes(x = study_hours, y = mean_lwr), linetype = "dashed") +
  geom_line(data = bands, aes(x = study_hours, y = mean_upr), linetype = "dashed") +
  geom_line(data = bands, aes(x = study_hours, y = pred_lwr), linetype = "dotted") +
  geom_line(data = bands, aes(x = study_hours, y = pred_upr), linetype = "dotted") +
  labs(
    title = "Figure 6. Mean 95% CI (dashed) and 95% Prediction Band (dotted)",
    x = "Study Hours (X)",
    y = "Grade (Y)"
  )
print(p6)
save_plot(p6, "Fig6_MeanCI_vs_PredictionBand.png")

# ============================================================
# FIGURE 7 (Optional): Boxplot of Grades (outliers + spread)
# ============================================================
p7 <- ggplot(df, aes(y = grade)) +
  geom_boxplot() +
  labs(
    title = "Figure 7. Boxplot of Grades",
    y = "Grade"
  )
print(p7)
save_plot(p7, "Fig7_Boxplot_Grades.png")

cat("\nAll figures saved in folder:", out_dir, "\n")
cat("PNG files are ready to insert into Word.\n")