Graphics Reference
In-Depth Information
The correlations between two time series can be computed using the
correlation function
correl()
in Excel, using statistical software, or
programmatically (for example, using Python). These correlation
relationships can be useful. For example, an investor may like the price
pattern of a particular stock (say, a tobacco stock) but would prefer to invest
in an alternative stock that has a similar price pattern to the original—that
is, a highly correlated stock. Real-world portfolio managers are interested in
correlations—they want diversified portfolios where the stocks they own are
not strongly correlated so that if the performance of one stock goes down,
the other stocks do not follow.
Note that any pair of time series can be transformed into a correlation
such as Google searches over time or time series of news story topics. In
the Supplementary Materials on this topic's companion website, see the
Stocks
example, which shows how to transform raw time series data into
links based on correlations.
Two Data Types (for Example, Board Memberships)
A
bipartite graph
has two different types of nodes, with linkages between
the different types. For example, a graph analysis of executives and their
board memberships reveals the connections between companies via board
members. The two different data types in this example are people and
companies. These are the nodes. The board memberships are the links that
connect a person to a company:
Exec, Board, Tenure
Sergey Brin, Google, 13 years
Paul Otellini, Intel, 11 years
Paul Otellini, Google, 9 years
...
Many Data Types (for Example, Social Links)
The idea of two data types can be extended to many different types of
data. People can be connected through many kinds of commonalities—for
example, LinkedIn builds connections via companies, friendships,
educational institutions, group memberships, and so on. In many business
cases, each type of connection may be in different databases, making the
integration of this disparate data much more difficult. Be sure to keep the
