598 Chapter 10. Software for Data Analysis
Figure 10.1.22: Power spectral density of the data shown in Fig.10.1.21.User interface.Of course the most important amongst these are the data analysis routines, which
depend on the particular methods and algorithms involved. Therefore during the
development of any data analysis package most attention should be paid to develop-
ment and implementation of the analysis methods, something that is unfortunately
sometimes overlooked. As explained above, the smart thing to do is to incorporate
the well tested routines into the code. But if the situation warrants design and
development from scratch, then utmost care should be taken in not only designing
the algorithms but also in implementing it.
In the following sections we will spend some time discussing the different parts
of a typical data analysis package.
10.2.ADataImport/ExportRoutines
Such routines can be fairly easily developed in any modern high level programming
language. Different languages provide the developer with different ways to handle
input/output. For example, C++ has input/output streams that can import and
export data in any format.
An important aspect of any data acquisition and storage system is the format of
the data file it generates. Since the data analysis program has to read data from
such files, one should be careful in choosing their format. The choice depends on
many factors but perhaps the most important is the size of the dataset. For small
data files, of the order of a few megabytes, one is better off writing data in delimited
ASCII test format. The reason is that it becomes fairly easy for the user to visually