12.5 ⁑Plots

If you have a large amount of data, you may want to consider using a mathematical tool (such as Matlab, Octave or GnuPlot) to generate your graphs as image files. If you use TeX, you may have excessively long document build times. Don't expect TeX to be able to compute logarithms or exponentials with the speed or precision of a custom-built numerical computing engine. Additionally, some packages can't parse scientific notation.

There are, however, some packages that use PostScript rather than TeX to perform the calculations (in which case you can't directly use PDFLaTeX) and some of them rely on TeX's shell escape to call an application, such as GnuPlot, to generate the drawing code (in which case you need the shell escape enabled). If you have applications such as GnuPlot or Matlab installed or if you are happy to use the latex+dvips+ps2pdf route to generate your PDF files, there are a number of useful packages listed on the graphics-plot topic. In addition, pgf/ tikz comes with an impressive mathematical engine provided by the pgfmath package. If you want to annotate your plot with text that matches your document format and have the patience to wait for your document to compile, you may be surprised by the images that can be produced using some of the packages available on CTAN.

Since the aim of this book is to be as useful to as many readers as possible, I'm again going to choose device-independent options that don't require any additional applications. (This is, after all, a book on administrative work not high performance computing or advanced mathematics.) As with pie charts and bar charts, the datatool package also provides a package (dataplot) for drawing plots from data stored in a database. However, if you don't intend using that data anywhere else in the document, I suggest you use a more flexible package such as pgfplots, which is described below. The other advantage of pgfplots over dataplot is that pgfplots can parse scientific notation and can cope with larger values. The remainder of this section discusses the pgfplots package.

The pgfplots package [26] uses pgf/tikz so make sure you have an up-to-date version of the pgf package installed. The pgfplots manual [26] is 500 pages long at the time of writing, so this section is a very brief introduction. See the user manual for further details.

Options can be set using

\pgfplotsset {⟨options⟩}

where ⟨options⟩ is a key=value list of options. The pgfplots manual recommends setting the compat key to benefit from recent features and to avoid possible changes if you recompile your document at a later date with a newer version. The log file will provide a suggested value.

Other common options include:

width: Sets the width of the final picture. An empty values indicates the default width or rescale in proportion to the height (so that the aspect ratio is maintained if you have set the height).
height: As above but for the height.
scale only axis: This is a boolean key. If true, the above width or height settings apply to the size of the axes rather than the overall picture size (which may include axis labels or tick marks).
xlabel: The x-axis label.
ylabel: The y-axis label.
title: The plot title.
major tick length: The length of major tick marks.
minor tick length: The length of minor tick marks.
tick align: The value may be one of: inside (default), outside or center. This indicates the location of the tick marks relative to the axis line.

There are many different types of plots and axes, but this introduction will just consider normal two-dimensional Cartesian plots, which can be produced using the axis environment:

\begin {axis}[options]
⟨plot commands⟩
\end {axis}

This should be placed inside a tikzpicture environment. The contents of the axis environment may contain one or more instance of

\addplot [⟨plot options⟩] ⟨plot specs⟩;

There are a number of different ways of specifying the plot, such as:

\addplot [⟨plot options⟩] coordinates {⟨co-ordinate list⟩} ⟨trailing path specs⟩;

where ⟨co-ordinate list⟩ is a space-separated list of co-ordinates. A standard Cartesian co-ordinate can be specified in the form (⟨x⟩,⟨y⟩), for example, (0.5,3.1). The ⟨trailing path specs⟩ are optional and are as the path specifications used by tikz's \path and \draw commands.

\addplot [⟨plot options⟩] table [column selection] {⟨file or inline table⟩} ⟨trailing path specs⟩;

where ⟨file or inline table⟩ is either a filename or the data arranged in the tabulated form of a space-separated data file. The ⟨column selection⟩ may contain the keys x=⟨column key⟩, to indicate the column containing the x values, and y=⟨column key⟩, to indicate the column containing the y values. The table data should include a header row at the start that may be used to reference the columns. If there is no header row, you can use header=false within ⟨column selection⟩ and use x index=⟨index⟩ and y index=⟨index⟩ instead of x and y to reference the required columns. The ⟨index⟩ should be an integer starting from 0 (the first column).

There are other \addplot specifications for evaluating expressions or using TeX's shell escape mechanism. See the pgfplots manual [26] for further details.

The ⟨plot options⟩ can be any of the path drawing options that are accepted by \path (or \draw), such as:

mark: Sets the marker style. There are three standard markers: * (filled circle), x (cross) and + (plus). Additional markers are available with tikz's plotmarks library (which can be loaded using \usetikzlibrary).
no markers: Overrides any mark value.
mark repeat: The value of this key should be an integer ⟨n⟩, to indicate that only every ⟨n⟩th mark should be drawn.
mark size: The size of the markers.
dashed: Dashed line style.
dotted: Dotted line style.
dashdotted: Dash-dot line style.
thin: Thin line width.
thick: Thick line width.
color: The value should a colour name used for stroking and filling. You can omit the color= and just write the colour name.

See the pgf manual [102] for further details.

Example 64. A Sample Plot (pgfplots package)

Suppose instead of the bar chart in Exercise 32 I want a graph of company profits where the x-axis represents the year and the y-axis represents the profit. I could have a file containing that data called, say, profits.dat that contains:

year profit
2010 52000
2011 50000
2012 75000
2013 60000

In which case, I can plot the data using:

\addplot table {profits.dat};

or I can just have the data inline as in the example document below:

\documentclass{article}
\usepackage{pgfplots}
\begin{document}

\begin{tikzpicture}
\begin{axis}
\addplot table
{
  year profit
  2010 52000
  2011 50000
  2012 75000
  2013 60000
};
\end{axis}
\end{tikzpicture}
\end{document}

This produces the following message in the transcript:

Package pgfplots Warning: running in backwards compatibility mode
(unsuitable tick labels; missing features). Consider writing
\pgfplotsset{compat=1.11} into your preamble.

So I need to add

\pgfplotsset{compat=1.11}

to the preamble. This produces the graph shown in Figure 12.12, which doesn't look right. The x-axis is far too cramped and doesn't need so many tick marks.

**Figure 12.12:** A Plot (First Attempt)
⇩

The positions of the x-axis tick marks can be changed using the xtick key. This may be set to \empty (generate automatically), the keyword data (use the co-ordinates provided by the first plot), or a comma-separated list of co-ordinates. (There are also analogous ytick and ztick keys.)

The tick labels can be changed using various keys. First is the xticklabels key which takes a comma-separated list of labels, where each label corresponds to a tick position. Alternatively, you can use xticklabel where the value is the code to produce the label. You can access information about the current tick using the following:

\tick

This is the current element of the tick option.

\ticknum

This command is only valid if the x tick label as interval option has been set (or y tick label as interval for the y-axis ticks) in which case it will contain the position of the next tick position.

The default label format is \pgfmathprintnumber{\tick}, which uses the number pretty-printing command provided by the pgfmath package. This is why the x-tick labels in Figure 12.12 have commas in them (for example, 2,010 rather than 2010). The comma can be removed by first using:

\pgfkeys{/pgf/number format/set thousands separator={}}

Alternatively, if you just want to change the style for a particular axis:

\pgfplotsset{x tick label style={
  /pgf/number format/set thousands separator={}
}}

The modified document is shown below. I've also added a plot title using title and axes labels using xlabel and ylabel:

\documentclass{article}

\usepackage{pgfplots}

\pgfplotsset{compat=1.11}

\begin{document}

\begin{tikzpicture}
\begin{axis}[xtick=data,% get x tick marks from data
 xlabel=Year,ylabel={Profits (\pounds)},% axes labels
 x tick label style={% change x tick label style
  /pgf/number format/set thousands separator={}%
 },
 title={Profits Since 2010},% plot title
 tick align=outside% ticks on the outside
 ]
\addplot table
{
  year profit
  2010 52000
  2011 50000
  2012 75000
  2013 60000
};
\end{axis}
\end{tikzpicture}

\end{document}

This produces the plot shown in Figure 12.13, which is much less cramped.

**Figure 12.13:** A Plot (Second Attempt)
⇩

In order to save space, the y-tick labels have been scaled by 10⁴ but since, in this case, the y axis represents profits it would look better if this scaling wasn't applied. There are a number of different ways of changing the scaling (see the manual [26]) but to switch it off for all axes, you just need the option scaled ticks=false. It's also possible that you don't like the boxed axes, but this can be changed using the option axis lines*=left. These extra options produce the plot shown in Figure 12.14.

**Figure 12.14:** A Plot (Final Attempt)
⇩

You can download or view this example where I have additionally loaded the plotmarks library and used the optional argument of \addplot to set the plot marks to filled diamonds (mark=diamond*) with the marker size set to 5 pt (mark size=5pt) and a thick cyan line stroke. (Try it for yourself before you download the example, as an additional exercise.)

⇦

⇧

⇨

This book is also available as A4 PDF or 12.8cm x 9.6cm PDF or paperback (ISBN 978-1-909440-07-4).