Track 3 · Visualising and Exploring Data · Lesson 9 of 11
Correlation heatmaps
Read a correlation matrix and its heatmap — sign, strength, redundancy — choose an honest colour scale, and recognise what a single correlation per pair cannot show.
Before building a price model, a valuation team wants a one-screen overview of how six property variables relate to each other. A correlation heatmap gives exactly that — and it is one of the easiest charts to over-read, because each coloured square compresses a whole scatter plot into one number.
A correlation matrix lists Pearson's correlation for every pair of numeric variables. Row , column holds between variable and variable . Three facts make it easy to read:
- The diagonal is all 1s: every variable correlates perfectly with itself.
- It is symmetric: the entry for (area, price) equals the entry for (price, area), so only one triangle carries information.
- Each entry is between and : the sign gives the direction, the size the strength of the linear association.
A heatmap colours each cell by its value, so that patterns in a large matrix jump out.
Included with a Formula Quest membership
It has 8 graded exercises with feedback, hints and worked solutions. After it you will be able to:
- Read a correlation matrix: the diagonal, symmetry, sign and strength
- Spot the variables most related to an outcome and the redundant pairs among predictors
- Use a symmetric diverging colour scale fixed at −1 to +1, with values printed
- Check scatter plots behind important cells: r misses curves and does not show causes