Pandemonium

Task: Easy

Objective:

Explore different clustering solutions in the Pandemonium app by adjusting clustering parameters and comparing the results.

Steps:

1. Install Pandemonium from GitHub:

   # install.packages("devtools")
    devtools::install_github("uschiLaa/pandemonium")
    library(pandemonium)
    pandemonium(b_anomaly$pred, b_anomaly$covInv, b_anomaly$wc, b_anomaly$exp)

2. Adjust Clustering Settings:
   • Setting A: Complete linkage, Canberra distance, 2 clusters, Coordinate- pull, Co-variance - included.
   • Setting B: Average linkage, Minkowski distance, 8 clusters, Coordinate- p-val, Co-variance - excluded.
3. Compare Results:
   • Setting A provides broad clusters, while Setting B reveals finer substructures.
   • Heatmap and dendrogram differences highlight how clustering changes based on parameters.

---

Task: Medium

For the medium-level task,we use detourr R package to generate two different tour paths of flea dataset projections and show both tours side-by side with linked brushing between the two displays.

Visualization of Linking & Brushing

We display two different 2D projections of the dataset using grand tour and little tour techniques.

Watch the video