added lab 4

2025-10-31 17:55:13 -04:00
parent dc2ceac7de
commit 88f2975b86
66 changed files with 730 additions and 301 deletions
@@ -0,0 +1,89 @@
+library(markdown)
+
+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")
+
+md <- c(
+	"# exploratory data analysis and models on the epi dataset",
+	paste0("date: ", as.character(Sys.Date()), "  "),
+	"",
+	"## dataset and choices",
+	# I regret my life choices
+	paste0("- **file**: `", basename(ctx$data), "`"),
+	paste0("- **region column**: `", ctx$region_col, "`"),
+	paste0("- **response var**: `", ctx$response, "`"),
+	paste0("- **regions**: `", ctx$region_a, "` vs `", ctx$region_b, "`"),
+	"",
+	"## 1) variable distributions",
+	"### 1.1 boxplots and histograms (with density!)",
+	paste0("![](", ctx$box_a, ")"),
+	paste0("![](", ctx$box_b, ")"),
+	paste0("![](", ctx$hist_a, ")"),
+	paste0("![](", ctx$hist_b, ")"),
+	"",
+	"### 1.2 qq plot (two-sample)",
+	paste0("![](", ctx$qq_fig, ")"),
+	"",
+	"## 2) linear models"
+)
+
+# Normalize data.frame because NOTHING WORKS
+row_list <- function(x) {
+	if (is.null(x)) return(list())
+	if (is.data.frame(x)) {
+		lapply(seq_len(nrow(x)), function(i) as.list(x[i, , drop = FALSE]))
+	} else if (is.list(x)) {
+		x
+	} else {
+		list()
+	}
+}
+
+if (!is.null(ctx$ols) && length(ctx$ols)) {
+	for (m in row_list(ctx$ols)) {
+		md <- c(md,
+		paste0("### ", m$name),
+		sprintf("- **r²**: %.4f  |  **aic**: %.2f  |  **bic**: %.2f", m$r2, m$aic, m$bic),
+		""
+		)
+	}
+}
+
+md <- c(md,
+	"### 2.2 same models on one region (comparison)",
+	if (!is.null(ctx$best_region_note)) ctx$best_region_note else "no note available.",
+	"",
+	"## 3) classification (knn, label = region)"
+)
+
+if (!is.null(ctx$knn) && length(ctx$knn)) {
+	for (k in row_list(ctx$knn)) {
+		md <- c(md,
+		paste0("### ", k$tag),
+		sprintf("- **k**: %d  |  **accuracy**: %.4f  |  **test n**: %d",
+				k$k, k$accuracy, k$n_test),
+		paste0("variables: `", paste(k$vars, collapse = ", "), "`"),
+		paste0("![](", k$confusion_fig, ")"),
+		""
+		)
+	}
+}
+
+# I hate markdown sometimes man
+md <- gsub("/home/ion606/Desktop/Homework/Data Analytics/Assignments/Assignment II/output/", "", md)
+
+# writeLines(md, "/home/ion606/Desktop/Homework/Data Analytics/Assignments/Assignment II/output/report.md")
+# writeLines(jsonlite::toJSON(ctx, pretty = TRUE, auto_unbox = TRUE),
+#            file.path(ctx$stats_dir, "summary.json"))
+
+md_file <- "output/report.md"
+html_file <- "output/report.html"
+pdf_file <- "output/report.pdf"
+
+setwd("/home/ion606/Desktop/Homework/Data Analytics/Assignments/Assignment II/")
+markdownToHTML(
+  md_file,
+  html_file
+)
+
+message("done")