When creating a visualization, there is a number of good practices you have to take into account. Essentially, the scope of any visualization is to provide an insight into a particular set of data. On top of that, it's supposed to be a summary of the data that quickly answers the main question of the analysis while also providing key metrics.
Let's take a look at the following Makeover Monday data set (which you can find it here ). We can also see the original dashboard attached to the dataset - make sure to check it out before reading the rest of this post..
Let's try to understand this visualization.. Well, I think you would agree with me that at a first glance it's pretty impossible to do so.
The dashboard aims to look at the Ecological Footprint of Countries in 2013 in relation to 5 different metrics. When you have a large list of countries, the use of bar charts should be avoided as you would end up with either a scroll bar (if you choose to visualize the countries vertically) or overlapped labels (see example above). You could instead opt for a drill down bar chart (from region for example) or an analysis on the top n countries.
When it comes to the type of chart, in this case a stacked bar chart is definitely not the greatest choice.. First of all, there are 5 different metrics used in the analysis, which means 5 different colors shown in the chart. Sure, this could work at times depending on the case and data set; However, in this case the colors look harsh and make the chart hard to read. What's more, while you can compare the proportion of each metric within each country, it becomes difficult to do so across same metrics for different countries; And of course, as the bars get smaller and smaller towards the right of the viz, it becomes impossible to read or differentiate anything..
Some thing to have in mind when creating a viz:
- Keep things simple! - Try to keep the number of distinct colours as low as possible as it would be much easier for the reader to understand the analysis.
- Choose the right type of chart
- Formatting
- Don't overlap data (labels)