This week I was given an excel spreadsheet of values and had to use Plotly to create a visualization of the data. The result I came up with is the above graphic. I plotted the data as time series data and shaded the area under the line to make it easier to see the change in the data. To implement the Part to Whole methodology with this time series data I re-expressed the high and low ends of the data in the two subplots and hopefully solve some issues with scaling. Because the data covers a broad range of values, and the changes at the top and bottom end are very subtle, it is easier to see those changes in the time series data by zooming in on the data.
Part to Whole is not perfect, nor is it best for every data set. The places this design framework works best are when comparing ratios or categorical data. In this example the time series is not really a collection of ratios nor is it categorical, and as such it cannot take full advantage of the design framework. Part to Whole can also limit the amount of data that can be presented in one graphic. For example a pie chart or bar plot becomes useless with too many categories or variables. Despite its limitations, the main idea of Part to Whole of breaking the data down into parts and looking for patterns in the parts still remains, and sometimes you can find patterns in those parts that you may have missed when analyzing the data in aggregate.