Key decisions when making heatmaps

In a previous post, I discussed several chart forms that can be used to illustrate time-series data with seasonality and trending. The post was inspired by a blog post at the Energy Institute at Haas. I held off on discussing one of the more interesting alt versions of the visualization. Here it is:The author experimented with this heatmap. The monthly dataset is laid out as a grid, with years as columns, and months as rows. Each cell is a specific month-year.They chose a color scheme that is quite easy to interpret. Our eyes quickly finds the orange-red cluster of values on the top right corner.The heatmap can be regarded as an annotated data table. Building heatmaps from numeric data presents several formidable challenges. Let's face them head on.What to encode with colors?Compared to other options such as bar lengths, positions of dots and lines and even pie angles, color encoding is less precise.Recall my self-sufficiency test. There is a reason why the entire dataset is printed…