Data—Collect, Clean, and Connect - Graph Analysis and Visualization - page 98

Matrix layouts are used by some software where all the possible links
are expected. For example, the d3.js chord diagram example uses a
two-dimensional (2D) matrix of data to define the links.
Putting It All Together
Theessentialfirststepsingraphanalysisandvisualizationaretoacquirethe
right data and to answer the objective. To recap, following are the key steps
to getting data:
• Objective —Prior to collecting data, ensure that the objective is known.
This will help establish what data needs to be collected and how it can
be prepared. For example, the question of whether to maintain multiple
links between nodes or aggregate them into a single link must be
understood in the context of the overall project goals.
• Collect —Graph data can exist in data in many different ways. It may be
necessary to collect data from multiple sources and/or evaluate the data
to determine if there is a way that links could be resident in the data and
potentially extracted (for example, by spreadsheets or
programmatically).
• Clean —Unfortunately, real-world data may often have quality issues,
and for graphs, you have additional concerns to address. Some graph
software cannot handle self-loops, duplicate links, missing nodes, or
Null values.
• Connect —Data must be transformed into a format that graph software
can use—typically a list of nodes and links. Once prepared, the data
must be output, and you have various file format alternatives. CSV,
GDF, GML, and JSON can all be straightforward to use when preparing
spreadsheets or via programming.
Summary
In any kind of data analysis or data visualization, data is required. It is
quitefeasibleforaprojecttofailwithinadequatedataandpoorpreparation.
Therefore, take care to identify the right data and prepare it appropriately.
With a well-defined objective and the right data, you can proceed to the next
step—graph analysis and visual layout, as described in Chapter 4. Finally,