Heatmaps (heat maps) are graphs produced by applying a color map to a 2D array of numbers. The resulting graph gives the viewer a quick, big-picture view of often complex relationships, making the heatmap an ideal tool for publication and presentation.
A correlation matrix is produced by repeating measurements across a fixed set of variables, on a group of subjects, then making pairwise comparisons between variables to determine their degree of correlation. A correlation matrix heatmap — or simply “correlation plot” — is produced by applying a color map to the correlation matrix. The resulting graph allows the viewer to quickly assess the degree of correlation between any two variables.
Heatmaps were supported in Origin 2019 but OriginPro 2020b has some new options added specifically for creating and customizing correlation plots. You can make a basic block-style correlation plot using tools built into Origin 2020b; or you can install a free Correlation Plot App and make some really jazzy correlation plots that incorporate such things as scatter points of varying size, oriented ellipses to denote positive or negative correlation, or labels of significance or insignificance marks, p-values, etc.
Before reviewing all of the creation and customization options, I want to point out how easy it is to create a basic correlation plot in Origin. If your subjects are arranged by row with your variables across columns, you can create a plot in two quick steps — no programming needed.
- Select your column variables and choose Statistics: Descriptive Statistics: Correlation Coefficient. Set your analysis and output options, including your correlation tests and click OK.
- Select the correlation matrix that is produced and choose Plot: Contour: Heatmap or Heatmap with Labels.
New Fill and Label Options for a Basic Heatmap
Grid-drawing Options: The first new Plot Details option we’ll mention is the addition of a Fill Display drop-down list to the Colormap tab. This gives you control over color fill with options for None (labels only), Square (all cells filled), Upper Triangle, Upper Triangle without diagonal, Lower Triangle, and Lower Triangle without diagonal.
Labeling Options: Origin includes templates for both labeled and unlabeled heatmaps. However, converting one to the other or changing the way in which labels display, is simply a matter of editing Plot Details Label tab controls.
- Clear or check the Enable box to show or hide labels.
- Set Label Display to Square, Upper Triangle or Lower Triangle.
- Set font, font color, label orientation, and other style options using the Font controls. Note that color = Auto automatically adjusts font color for maximum contrast with cell background color.
- Use the Numeric Display Format to modify the display of numbers. For instance, in the above figure, we enter .2 to display correlation numbers to two decimal places (for more information, see Custom Numeric Formats in the Origin Help file).
Free Add-on: Correlation Plot App
For creating correlation plots beyond the basic block-style heatmap, OriginLab is offering a free Correlation Plot App. You can install this App via your software’s App Center by pressing F10 and searching for “Correlation Plot”; or you can download the App directly from the OriginLab File Exchange and drop it into your Origin workspace. If you don’t yet own Origin 2020b, you can still give the Correlation Plot App a test-drive by installing the Trial Version of OriginPro 2020b. The Trial Version lets you evaluate OriginPro free-of-charge for 21 days.
The Correlation Plot App has a simple, interactive interface that allows you to preview the resulting graph as you go. Note, too, that you can tweak the resulting graph via Plot Details or Color Scale Control dialog.
The following correlation plots were all created using the Correlation Plot App. They can be found in your software’s Learning Center (Help: Learning Center or press F11 and search on “correlation”). Each plot is accompanied by a sample dataset and basic instructions for creating the plot.
We hope that this has been informative. If you routinely work with multivariate data, we hope we’ve piqued your interest enough to check out OriginPro 2020b’s easy-to-create Correlation Plots. Thanks for reading. Look for them in the upcoming release of Origin 2020b, due out some time in April of 2020.