Heatmaps: A Multivariate

Visualization Method

April 5, 2017

Why use heatmaps

• Matrix visualized with colour gradients

• Visually recognize patterns in data

• Condense multiple response and • Condense multiple response and

predictor variables into one figure

• Highlight similarities and/or

differences between predictor and

response variables

History of heatmaps

Creating heatmaps in R

1. Create data matrix.

2. Scale the data.

3. Create distance values – dist().

o euclidean, maximum, manhattan, canberra, binary or minkowski

4. Cluster the values by creating a dendrogram – hclust().4. Cluster the values by creating a dendrogram – hclust().

o ward.D, ward.D2, single, complete, average, mcquitty, median, centroid

• Heatmap -> heatmap.2

• Aheatmap -> pheatmap

• Heatmap3

• Heatmaps in ggplot2 (not as good)

References

Wilkinson, L. and M. Friendly. 2009. The History of the Cluster Heat Map. The

American Statistician 63(2):179-184. DOI: 10.1198/tas.2009.0033.