added assignment II

2025-10-13 12:44:58 -04:00
parent df21d7f281
commit 8b274f6bd6
36 changed files with 1126 additions and 1 deletions
@@ -0,0 +1,59 @@
+source("/home/ion606/Desktop/Homework/Data Analytics/Assignments/Assignment II/R/00_utils.R")
+ctx <- jsonlite::fromJSON("/home/ion606/Desktop/Homework/Data Analytics/Assignments/Assignment II/output/ctx.json")
+df <- suppressMessages(readr::read_csv(ctx$data, show_col_types = FALSE))
+
+fit_ols <- function(data, y_col, x_cols, name_tag, fig_dir, stats_dir) {
+	d <- data |> dplyr::select(all_of(c(y_col, x_cols))) |> tidyr::drop_na()
+	for (xc in x_cols) d[[xc]] <- tf_pos(d[[xc]])
+	f <- as.formula(paste(y_col, "~", paste(x_cols, collapse = " + ")))
+	m <- lm(f, data = d)
+
+	# residuals vs fitted
+	res_path <- file.path(fig_dir, paste0("residuals_", sanitize(name_tag), ".png"))
+	p_res <- tibble(fitted = fitted(m), resid = resid(m)) |>
+		ggplot(aes(fitted, resid)) +
+		geom_point(size = 1.6) +
+		geom_hline(yintercept = 0) +
+		labs(title = paste("residuals vs fitted:", name_tag),
+			x = "fitted", y = "residuals") +
+		theme_minimal()
+	save_plot(p_res, res_path)
+
+	# scatter first predictor vs response
+	first <- x_cols[1]
+	sc_path <- file.path(fig_dir, paste0("scatter_", sanitize(name_tag), "_", sanitize(first), ".png"))
+	p_sc <- d |>
+		ggplot(aes(.data[[first]], .data[[y_col]])) +
+		geom_point(size = 1.6) +
+		labs(title = paste(first, "vs", y_col), x = first, y = y_col) +
+		theme_minimal()
+	save_plot(p_sc, sc_path)
+
+	# summary to text
+	summ_path <- file.path(stats_dir, paste0("ols_", sanitize(name_tag), ".txt"))
+	capture.output(summary(m), file = summ_path)
+
+	gl <- broom::glance(m)
+	list(
+		name = name_tag,
+		rsq = unname(gl$r.squared),
+		aic = unname(gl$AIC),
+		bic = unname(gl$BIC),
+		nobs = stats::nobs(m),
+		summary_file = summ_path,
+		residuals_fig = res_path,
+		scatter_fig = sc_path
+	)
+}
+
+ols <- list()
+if (length(ctx$predictors)) {
+	for (p in ctx$predictors) {
+		tag <- paste0("full: ", ctx$response, " ~ ", paste(p, collapse = " + "))
+		ols <- append(ols, list(fit_ols(df, ctx$response, p, tag, ctx$fig_dir, ctx$stats_dir)))
+	}
+}
+
+ctx$ols <- ols
+writeLines(jsonlite::toJSON(ctx, pretty = TRUE, auto_unbox = TRUE), "/home/ion606/Desktop/Homework/Data Analytics/Assignments/Assignment II/output/ctx.json")
+message("ols (full) done")