test

Author: Senomi-Jayasinghe

PCA & Kmeans Clustering - 31376.R

@@ -0,0 +1,38 @@
+data <- read.csv("Medical_Dataset_with_Disease_Types.csv")
+
+data1 <- data[, 1:10]
+head(data1)
+
+install.packages("factoextra")
+library(factoextra)
+
+pca <- prcomp(data1, center = TRUE, scale. = TRUE)
+head(pca)
+
+#SCREE PLOT
+fviz_eig(pca, addlabels = TRUE, ylim = c(0,60))
+
+fviz_pca_biplot(pca,
+                repel = TRUE,
+                col.var = "blue",
+                col.ind = "gray40")
+
+pca_scores <- as.data.frame(pca$x)
+
+#let's use first 4 pc for clustering
+pc_data <- pca_scores[, 1:4]
+head(pc_data)
+
+#K means library
+library(dplyr)
+
+set.seed(125)
+kmeans_result <- kmeans(pc_data, centers = 3, nstart = 25)
+
+#add cluster result to pca data
+pc_data$Cluster <- as.factor(kmeans_result$cluster)
+pc_data$True_label <- data$Disease_Type
+
+#performance
+table(Cluster = pc_data$Cluster,Actual_Disease = pc_data$True_label)
+