Solution

Gnuplot is a very popular software package used to create charts based on mathematical functions and data points. You can use Gnuplot in a stand-alone fashion or as an embedded viewer for graphs created by other applications. In this section, you will learn how to generate files that can be easily visualized using Gnuplot.

The first step in using Gnuplot is to make sure that it is properly installed in your system. You can easily install this package for data visualization by visiting its website (www.gnuplot.info) and downloading the required files. There are binary installation files available for most operating systems, including Windows, Mac OS X, and Linux. Run the installer and execute the main application. You should see something similar to the screen displayed in Figure 6-1.

You can see the results for this simple function plot in Figure 6-2.

In this plot, you supplied the mathematical function defined by sin(x) , and Gnuplot is responsible for creating a plot of the values, where the default range is from -10 to 10. You can easily tweak the parameters used to determine the range, as well as other attributes of the plot such as the title, the legend, and the units used in both axes.

Another way to use Gnuplot is to directly plot numeric data, instead of a mathematical function. This is possible by referencing the name of the files that should be imported by the plot command. Most data imported in this way is in the csv (comma-separated values) format, although Gnuplot doesn’t mandate that the number be in csv—any file with numeric data organized as columns will do.

The first command is used to tell Gnuplot that the data in the x axis is time-oriented. The second command defines the separator used in the file. The third command describes the date format stored in the csv file. Finally, the last line tells Gnuplot to plot the contents of file IBM.csv , using columns 1 and 7 (column 1 contains dates, while column 7 has adjusted closing prices), and with the title of the time series defined by the headers for each column of the csv file.

These commands will generate the plot displayed in Figure 6-3.

The plot in Figure 6-3 is just a sample of what Gnuplot can do. There are literally hundreds of parameters that can be tweaked using the set command . Among these options, you can find three-dimensional plots, different colors, and line styles, among others.

To solve the problem presented in this section, you need to create a class that receives some data in the form of vectors of doubles or strings and produces output data suitable for consumption by Gnuplot. I have created such a class, which is called GnuplotPlotter and is responsible for the generation of the files needed by Gnuplot.

To access the results of the class, two member functions are available. The csvWrite member function will write the data stored in GnuplotPlotter to the file specified in the constructor, using the csv format. The second member function is generateCmds , which allows one to create a command file with the necessary instructions to Gnuplot. This way, you don’t need to worry about the exact syntax for plotting the file. The commands are stored in a filename specified by the parameter cmdFileName .

An example for the GnuplotPlotter class is given in the main function. First, you need to define two vectors with the desired data. In this case, you will use data generated by the function sin, which returns the trigonometric sine of a number. Notice that we do this only to simplify data testing. The data file, however, can have numbers generated from any source. After the content has been defined, you can call the member functions csvWrite and generateCmds to create the files needed by Gnuplot. You can see the result of this process in Figure 6-4.

Complete Code

Running the Code

The plot will be displayed in a separate window, as shown in Figure 6-4.

Solution

Although it is great to have the ability to create charts with external packages such as Gnuplot, sometimes it is necessary to have a larger degree of control over the output generated by plots. If you cannot find a way to use one of the parameters in Gnuplot to get the desired results, it becomes necessary to implement a plotting solution that runs in C++. This section shows how to achieve this.

There are many graphical libraries available for C++ developers, and the final decision depends mostly on your target environment. However, in this section, I use the Qt library to implement the desired solution.

Qt is probably the easiest to use graphical programming package around. You will see that with just a dozen lines, we are able to create a complete application. Moreover, Qt is available for all major operating systems, so that your application can be easily ported to other targets as necessary.

The class used is called QtPlotter , and it receives data using the setData member function, just as we did with the GnuplotPlotter. The main part of the implementation, however, is performed in the PlotWindow class , which is derived from QMainWindow , one of the key classes in the Qt framework. The PlotWindow class is responsible for managing the window and, most important, painting the plot when necessary.

In the next step, the function paintEvent draws the tick markers along the axis. Finally, the code paints lines between the points given as input to the plot.

This code does most of what is necessary to create a Qt application and display a window on the screen. The window created is the PlotWindow class that we discussed previously, so that the plot is displayed as desired. The QApplication object is part of the Qt framework. It manages the workflow of a graphical application, including menus and windows. When creating a QApplication, we are able to determine the application name with the setApplicationName member function. Finally, we resize and show the plot window and call the exec member function to start the window display loop.

Figure 6-5 shows the results of this code.

Complete Code

Running the Code

To use the QtPlotter class, you need to have the Qt4 library installed in your system. The installation process requires you to visit the developer website (www.qt.io/developers/), download, and run the installer application. After the installation is complete, the libraries will be copied to a user-defined folder.