Materials for Applied Data Science profile course INFOMDA2 Battling the curse of dimensionality.

1 Introduction

In this practical, we will apply hierarchical and k-means clustering to two synthetic datasets. We use the following packages:


The data can be generated by running the code below.

1. The code does not have comments. Add descriptive comments to the code below.

sigma      <- matrix(c(1, .5, .5, 1), 2, 2)
sim_matrix <- mvrnorm(n = 100, mu = c(5, 5), 
                      Sigma = sigma)
colnames(sim_matrix) <- c("x1", "x2")

sim_df <- 
  sim_matrix %>% 
  as_tibble() %>%
  mutate(class = sample(c("A", "B", "C"), size = 100, 
                        replace = TRUE))

sim_df_small <- 
  sim_df %>%
  mutate(x2 = case_when(class == "A" ~ x2 + .5,
                        class == "B" ~ x2 - .5,
                        class == "C" ~ x2 + .5),
         x1 = case_when(class == "A" ~ x1 - .5,
                        class == "B" ~ x1 - 0,
                        class == "C" ~ x1 + .5))
sim_df_large <- 
  sim_df %>%
  mutate(x2 = case_when(class == "A" ~ x2 + 2.5,
                        class == "B" ~ x2 - 2.5,
                        class == "C" ~ x2 + 2.5),
         x1 = case_when(class == "A" ~ x1 - 2.5,
                        class == "B" ~ x1 - 0,
                        class == "C" ~ x1 + 2.5))

2. Prepare two unsupervised datasets by removing the class feature.

3. For each of these datasets, create a scatterplot. Combine the two plots into a single frame (look up the patchwork package to see how to do this!) What is the difference between the two datasets?

2 Hierarchical clustering

4. Run a hierarchical clustering on these datasets and display the result as dendrograms. Use euclidian distances and the complete agglomeration method. Make sure the two plots have the same y-scale. What is the difference between the dendrograms? (Hint: functions you’ll need are hclust, ggdendrogram, and ylim)

5. For the dataset with small differences, also run a complete agglomeration hierarchical cluster with manhattan distance.

6. Use the cutree() function to obtain the cluster assignments for three clusters and compare the cluster assignments to the 3-cluster euclidian solution. Do this comparison by creating two scatter plots with cluster assignment mapped to the colour aesthetic. Which difference do you see?

3 K-means clustering

7. Create k-means clusterings with 2, 3, 4, and 6 classes on the large difference data. Again, create coloured scatter plots for these clusterings.

8. Do the same thing again a few times. Do you see the same results every time? where do you see differences?

9. Find a way online to perform bootstrap stability assessment for the 3 and 6-cluster solutions.

4 Challenge question

10. Create a function to perform k-medians clustering

Write this function from scratch: you may use base-R and tidyverse functions. Use Euclidean distance as your distance metric.

Input: - dataset (as a data frame) - K (number of clusters)

Output: - a vector of cluster assignments

Tip: use the unsupervised version of sim_df_large with K = 3 as a tryout-dataset

11. Add an input parameter smart_init. If this is set to TRUE, initialize cluster assignments using hierarchical clustering (from hclust). Using the unsupervised sim_df_small, look at the number of iterations needed when you use this method vs when you randomly initialize.