renaming

2026-02-17 23:13:20 +03:00
parent 65218abfb1
commit e52dde575a
429 changed files with 875 additions and 14 deletions
--- a/science/r/4.rmd
+++ b/science/r/4.rmd
@ -0,0 +1,158 @@
+---
+title: "Lab4: K-means, DB-scan and dendrograms"
+author: "Vladislav Litvinov <vlad@sek1ro>"
+output:
+  pdf_document:
+  toc_float: TRUE
+---
+# Data preparationc
+```{r}
+setwd('/home/sek1ro/git/public/lab/ds/25-1/r')
+load("./income_elec_state.Rdata")
+df = income_elec_state
+df$incomelog = log10(df$income)
+remove(income_elec_state)
+```
+# Function to compute Within-Cluster Sum of Squares for choosing optimal K
+```{r}
+elbow_wss = function(df) {
+  max_k = 10
+  wss = numeric(max_k)
+  
+  for (i in 1:max_k) {
+    res = kmeans(df[,1:2], centers = i)
+    wss[i] = res$tot.withinss 
+  }
+  
+  plot(1:max_k, wss, type="b")
+  
+  wss_diff = diff(wss)
+  wss_ratio = wss_diff[-1] / wss_diff[-length(wss_diff)]
+  return(which.min(wss_ratio))
+}
+```
+# Scatter-plot: elec vs income and log-income
+```{r}
+library(ggplot2)
+
+plot_kmeans = function(data, k, log) {
+  res = kmeans(data, centers = k)
+  centers = as.data.frame(res$centers)
+  centers$cluster = as.factor(1:k)
+  data$cluster = as.factor(res$cluster)
+  data$state = rownames(df)
+  
+  plt = ggplot() +
+    geom_point(
+      data = data,
+      aes(
+        x = income,
+        y = elec, 
+        color = cluster
+      )
+    ) +
+    geom_text(
+      data = data,
+      vjust = 1.5,
+      size = 2,
+      aes(
+        x = income,
+        y = elec,
+        label = state
+      )
+    ) +
+    geom_point(
+      data = centers,
+      shape = 17,
+      size = 5,
+      aes(
+        x = income,
+        y = elec,
+        color = cluster,
+      )
+    ) +
+    theme_minimal()
+  
+  if (log) {
+    plt = plt + scale_x_log10() + scale_y_log10()
+  }
+  
+  print(plt)
+}
+
+data = df[,c("income", "elec")]
+datalog = df[,c("incomelog", "elec")]
+k = elbow_wss(data)
+klog = elbow_wss(datalog)
+plot_kmeans(data, k, log=FALSE)
+plot_kmeans(data, klog, log=TRUE)
+
+```
+# Map of USA
+```{r}
+library(maps)
+res = kmeans(data, centers = k)
+map_color = res$cluster[order(names(res$cluster))]
+map("state", fill = TRUE, col = map_color)
+
+Q1 = quantile(df$elec, 0.25)
+Q3 = quantile(df$elec, 0.75)
+IQR = Q3 - Q1
+min = Q1 - 1.5 * IQR
+max = Q3 + 1.5 * IQR
+df = subset(df,  df$elec > min & df$elec < max)
+
+data = df[,c("income", "elec")]
+datalog = df[,c("incomelog", "elec")]
+k = elbow_wss(data)
+klog = elbow_wss(datalog)
+plot_kmeans(data, k, log=FALSE)
+plot_kmeans(data, klog, log=TRUE)
+
+```
+# Differences within dendrogramm algo: 'single', 'complete', 'ward.D', 'average'
+```{r}
+
+library(ggdendro)
+
+
+plot_hclust = function(df, linkage, k) {
+  data = df[,c("income", "elec")]
+  distance = dist(data, method = "euclidean")
+  clust = hclust(distance, method = linkage)
+  data$cluster = as.factor(cutree(clust, k = k))
+  data$state = rownames(df)
+  print(cutree(clust, k = k))
+  print(data)
+  plt = ggplot() +
+    geom_point(
+      data = data,
+      aes(
+        x = income,
+        y = elec, 
+        color = cluster
+      )
+    ) +
+    geom_text(
+      data = data,
+      vjust = 1.5,
+      size = 2,
+      aes(
+        x = income,
+        y = elec,
+        label = state
+      )
+    )
+    theme_minimal()
+    print(plt)
+}
+
+plot_hclust(data, "average", 5)
+
+distance = dist(data, method = "euclidean")
+
+clust = hclust(distance, method = "single")
+plot(ggdendrogram(clust))
+
+cutree(clust, k = 3)
+```