Correspondence Analysis is a powerful technique for uncovering relationships between items and attributes, but its value depends heavily on how clearly the results are presented. A perceptual map crowded with overlapping labels, dense central clusters, and lengthy response names can obscure the very insights it is meant to reveal. That is why Halo Reports Correspondence Analysis comes with a rich set of presentation tools that let you refine the chart's appearance without compromising the underlying statistics. In this article, we will walk through these options, from automatic label placement and decluttering settings to point-level styling and reusable color palettes, so you can turn a raw analysis into a clear, presentation-ready visual that draws attention exactly where you want it.
If you want to know more about Correspondence Analysis, visit our Learning Center.
Halo Reports Correspondence Analysis provides several features aimed at improving the visual presentation, helping to create the chart in the best possible way. First of all, note that the automatic label relocator ensures that there is no overlap between the labels on the chart. Gentle leading lines connect the labels with the points if they have to be moved away from them to avoid conflicts. This makes it possible to follow the information even on charts with a lot of noise from many points in small clusters.
Furthermore, the chart exposes several options to declutter the chart, without impacting the analysis results. As mentioned before, changing the query definition by applying a subset will regenerate the whole statistical analysis with new input. In case you would prefer the model to consider all the data points in the calculation, but clean up the chart a bit to focus on specific pieces of information, there are some options on the right-side settings that can help you.
In the general settings, you can find the “Hide central points” option. It hides a given percentage of the points closest to the chart origin. That can be very helpful if you want to focus on analyzing the most distinctive data points, automatically decluttering the middle of the chart, formed with lots of less unique points. You can also remove the additional disclaimers (both the “tutorial” available by default and the contextual on-click one) to gain more space for the chart itself.
Advanced settings include Edit Items and Edit Attributes buttons, which will take you to a dedicated menu, allowing you to remove specific points from the chart. This window allows you to rename the specific responses, which can be useful to shorten the ones that decrease chart readability due to their length. That makes it possible to review the labels on a distinct chart without having to do it outside of Halo or during data preparation (although if you have a recurring challenge with longer labels, creating a permanent recode and using that on a chart instead of the original question may be a good idea).
Finally, you can enter the chart edit mode by pressing the brush icon in the toolbar at the top of the chart. In that mode, you can select a specific series, point, legend, or explainer box and impact its visual aspects. In all cases, you have control over the font displayed. For example, if your chart is crowded, you may want to decrease the font size of the data points for better readability. The data points also expose some control over their visual aspects. If you select a whole series, you can change the marker and color for the points in that series.
You can also give points in the same series different colors. Depending on the need, you can reach that result in two ways. If you’d like to differentiate each point in either attributes or item series, you can simply change the switch in the general settings. This will automatically change the color of the series to a default color palette. You can reuse an existing palette or create a new one according to your preferences. Read more about palettes HERE.
If, rather than making a fully distinct palette, what you are really looking for is to highlight one (or a few) main points of interest, it will be easier to use the chart edit mode to select a single point and change its color. Note that this effectively generates a palette, with this one entry having a fixed mapped color, and all the others having the previously selected global color. You can keep highlighting other points with the same or a different color. Since palettes are assigned to users, you can save the palette for further reuse, for example, if you keep highlighting the same brands - save the palette and simply reapply it on each instance of your correspondence analysis rather than manually clicking each point every time.