This is the corrected mode colors. Note: the colors on the markers on the graph are not the representative color of the clusters, they are just colors from the default set.

There was a mistake in the script for calculating the mode colors, please refer to the updated scatter plot.

I have explored many different chart types offered by Plotly such as line chart and basic histogram. However, line chart will get really messy when displaying so many event types. Similarly, basic histograms lay different colour event type bars side by side for 19 years resulting in a very large chart with tiny elements. By choosing stacked histograms, I saved 9 times the horizontal space since there are 9 event types of interest. In addition, I am still able to get exact figures of each event type of each year by pointing at the chart to reveal the hovering text offered by Plotly.

We have always been looking at a scatter plot of data from the top down perspective. If we were to do 2 dimensional scatter plot of violence against civilians in Somalia for 19 years, we would have to create 19 different maps for each year in order to study the trend and spatial pattern of violence against civilians’ events. However, we are able to create 3D scatter plots with Plotly that viewer can interact with by moving and rotating the scatter plot. It is as if we are layering many 2D scatter plots upwards to form a more interesting map. However without the interaction provided by the javascript behind Plotly, it would be difficult to study such complex structure.