Guidelines
Use When
- Representing the statistical correlations between pairs of variables in a set.
Don′t Use When
- Representing the distribution of items across two axis variables, absent of statistical correlation values.
- Displaying the distribution of a set of items relative to three variables, one of which is expressed through the size of the plotted item.
Visual Language
- Always use an even number of ranges within the legend scale, equally split between positive and negative correlation.
- The positive and negative correlation ranges each have a base color,
$mds-visualization-color-correlation-positive
and $mds-visualization-color-correlation-negative
to represent the range including the edge value (1 and −1 respectively). The subsequent ranges in each should utilize tints of the base color. These tints can be adjusted to match the distribution of the number of ranges, i.e., 25% tints could be used with 8 total legend items, while 30% may be more appropriate with 6 total legend items.
- Never assign new base colors to the Correlation Matrix. The base colors should remain consistent across implementations to help recognizability and understanding.
- Use mirroring to duplicate the data allowing for either axis as a starting point for comparison.
Responsiveness
- When the data cells become too small to contain the text, the cell displays only a color value.
Editorial
- Strive for short, succinct item names that clearly describe the data.
Code Reference
Necessary code documentation can be found in the MBC repository.