Playfair is a new project for exploring the features, pros, and cons of SharpPlot, and how it compares to the big popular projects like Vega-Lite, ggplot2 and plotly.
Playfair is also for exploring and what needs to be done to to cover SharpPlot to make it as easy and consistent to use as possible.
It should be noted the main concern is with statistics and business graphics, so the focus is on bar charts, line charts, and scatter plots. 3D plots and the like are not particularly relevant though SharpPlot clearly has some of that capability.
Also it is assumed that a sophisticated database and query language is available to produce the data ready to be plotted. There is little need for the chart library to summarize data, compute totals and averages, and do general analytics, even though SharpPlot and the rest can in fact do this to some degree. However, it can be useful to delegate splitting and grouping data to the chart library in certain circumstances.
There are three completely superficial (but to the casual observer often fatal) cons of SharpPlot which need to to be noted but dispensed with: the SharpPlot website/help, the Wizard provided in the Dyalog Windows IDE, and the default color scheme. The website and CHM help both display charts as fuzzy, out-of-focus, pixelated png files. (That's not good for a graphics package!) The wizard is unusable, the default color palette hideous.
The Grammar of Graphics
The primary difference between SharpPlot and Vega-Lite and ggplot2 is that the latter two are based on The Grammar of Graphics. This means things are ... abstract. To see how abstract, this quote from the home page of ggplot2 should suffice:
It’s hard to succinctly describe how ggplot2 works because it embodies a deep philosophy of visualisation. However, in most cases you start with ggplot(), supply a dataset and aesthetic mapping (with aes()). You then add on layers (like geom_point() or geom_histogram()), scales (like scale_colour_brewer()), faceting specifications (like facet_wrap()) and coordinate systems (like coord_flip()).
Now there is nothing wrong with deep philosophy, abstractions and generalizations, and for those who design and create charts for a living, or better yet, those who design systems that design charts, having the power this ultimately delivers is no doubt a good thing. Plotly and SharpPlot both fly a little closer to the ground, and provide specific functions that directly create bar charts and scatter plots and so on. Of course, given a grammar of graphics, the first thing we want to do is completely hide it from our end users, so it remains to be seen if it really is an advantage for basic business graphics.
One potential con of SharpPlot compared to the other packages (those based on GoG or not), is that the SharpPlot draw methods do not take a consistent specification for the data on the
y axes. For example, the SharpPlot
DrawBarChart method takes a vector of values for the quantitativive axis, usually
y. The categorical axis is not defined and must be manually labeled. To get a horizontal bar chart, the bar-chart-specific
Horizontal property must be specified, instead of just switching the specification of the data feeding the
y axes. A little more generalization would help here. Be that as it may, end users are probably more comfortable selecting chart type
BarChart and then clicking a check box for a
Horizontal option, without thinking much about axes.
In future posts we will be putting SharpPlot through its paces, comparing and contrasting with the other charting libraries.