Drawing a bar chat (horizontal) or log line chart

Question

I am trying to draw this graph, one is linear another is exponential and proportional to the size of the data. The numbers are so different from each other, I don't know how to make it more understandable.

Is it possible to make this bar chart horizontal or a log line chart?

\documentclass{article}
\usepackage[margin=0.5in]{geometry}
\usepackage{textcomp}
\usepackage{pgfplots}
\pgfplotsset{width=10cm,compat=1.9}
\begin{document}
\begin{tikzpicture}
    \begin{axis}[
            x tick label style={
                /pgf/number format/1000 sep=},
            xlabel=AST nodes,
            ylabel=Time in seconds,
            enlargelimits=0.05,
            legend style={at={(0.5,-0.1)},
                anchor=north,legend columns=-1},
            ybar interval=0.7
        ]
        \addplot 
        coordinates {(3,0.009)(30,0.003)(111,0.005)(354,0.019)(1083,0.097)(3270,0.044)(9831,0.064)(29514,0.501)(88563,2.276)(265710,7.439)(797151,27.578)(2391474,128.611)};
        \addplot 
        coordinates {(3,0.091)(30,0.495)(111,2.789)(354,3.390)(1083,5.021)(3270,20.149)(9831,48.015)(29514,158.442)(88563,857.381)(265710,2693.862)(797151,8771.571)};
        \legend{Static scheduling,Demand Scheduling}
    \end{axis}
\end{tikzpicture}
\end{document}

This is the source table:

\begin{figure}[htbp]
\begin{center}
\scalebox{0.9}{
\begin{tblr}
{
colspec      = {X[c,m]X[c,m]X[c,m]X[c,m]X[c,m]X[c,m]},
cell{1}{1}   = {r=2}{},
cell{1}{2}   = {r=2}{},
cell{1}{3,5} = {c=2}{},
cell{14}{5}  = {c=2}{},
hlines,
vlines,
}
AST depth & Nodes             & Visit sequence evaluation &  & Demand evaluation &    \\
& & ET in sec.                 & Memory in MB & ET in sec.                   & Memory in MB \\
1 & 3 & 0.009 & 2.971 & 0.091 & 7.741 \\ 
2 & 30 & 0.003 & 2.921 & 0.495 & 12.362 \\ 
3 & 111 & 0.005 & 3.646 & 2.789 & 21.288 \\ 
4 & 354 & 0.019 & 7.872 & 3.390 & 520.246 \\ 
5 & 1083 & 0.097 & 13.091 & 5.021 & 1016.131 \\ 
6 & 3270 & 0.044 & 46.332 & 20.149 & 1745.389 \\
7 & 9831 & 0.064 & 57.778 & 48.015 & 4564.360 \\ 
8 & 29514 & 0.501 & 155.338 & 158.442 & 7755.513 \\ 
9 & 88563 & 2.276 & 198.215 & 857.381 & 12081.491 \\ 
10 & 265710 & 7.439 & 940.484 & 2693.862 & 15053.265 \\ 
11 & 797151 & 27.578 & 1368.750 & 8771.571 & 15862.919 \\ 
12 & 2391474 & 128.611 & 3279.699 &  \text{JVM crashed}&   \\
\end{tblr}}
\end{center}
    \caption{Benchmarks of running \href{https://github.com/boyland/aps/blob/master/examples/nested-cycles.aps}{\texttt{nested-cycles}} example}
    \label{fig:nested-cycles-benchmark}
    The exponential nature of the demand schedule becomes evident as number of the AST nodes gets larger.
\end{figure}

Answer 1

I am unsure what you want to plot here: you have a set of wildly different numbers, and it is unclear to me which are the independent variables. So I will plot memory usage against the depth, but you can adapt it quite easily to other scenarios, I hope.

I would suggest using pgfplots, loading the data whole, and then exploring the different diagrams to find one that conveys what you need. In the following example, I load the data in-memory with pgfplotstable (notice: nan means "not a number", and is useful to represent missing data), and then I plot on a linear scale the memory of the two approaches with the depth as the independent variable:

\documentclass[border=10pt]{standalone}
\usepackage[T1]{fontenc}
\usepackage{pgfplots}\pgfplotsset{compat=newest}
\usepackage{pgfplotstable}
\pgfplotstableread{
depth  Nodes    St       Sm        Dt        Dm
1      3        0.009    2.971     0.091     7.741
2      30       0.003    2.921     0.495     12.362
3      111      0.005    3.646     2.789     21.288
4      354      0.019    7.872     3.390     520.246
5      1083     0.097    13.091    5.021     1016.131
6      3270     0.044    46.332    20.149    1745.389
7      9831     0.064    57.778    48.015    4564.360
8      29514    0.501    155.338   158.442   7755.513
9      88563    2.276    198.215   857.381   12081.491
10     265710   7.439    940.484   2693.862  15053.265
11     797151   27.578   1368.750  8771.571  15862.919
12     2391474  128.611  3279.699  nan       nan
}\data
\begin{document}
\begin{tikzpicture}[]
    \begin{axis}[
        axis line style = {thick, gray},
        enlarge x limits,
        enlarge y limits,
        xlabel = {Depth},
        % every axis x label/.append style = {below, gray},
        ylabel = {memory (MB)},
        legend style = {nodes={right, font=\scriptsize},
            at={(0.05,0.6)}, anchor=west},
        clip mode = individual,
        ]
        \addplot table[x=depth, y=Sm]{\data};
        \addplot table[x=depth, y=Dm]{\data};
        \legend{Static memory, Dynamic memory}
    \end{axis}
\end{tikzpicture}
\end{document}

...details like formats of numbers and labels can be adjusted later. Maybe a logarithmic y-axis is better? Add

 ymode = log,

(and adjust the legend position at 0.05, 0.9) and you have:

Maybe you want to add also the number of nodes for each run, and the information about the crash. This is a bit more complex, and I had to search a bit on this site, in the manuals of TikZ, and in the pgfplots/pgfplotstable's ones, but a proposal is:

\documentclass[border=10pt]{standalone}
\usepackage[T1]{fontenc}
\usepackage{pgfplots}\pgfplotsset{compat=newest}
\usetikzlibrary{shapes.symbols}% for starburst
\usepackage{pgfplotstable}
\pgfplotstableread{
depth  Nodes    St       Sm        Dt        Dm
1      3        0.009    2.971     0.091     7.741
2      30       0.003    2.921     0.495     12.362
3      111      0.005    3.646     2.789     21.288
4      354      0.019    7.872     3.390     520.246
5      1083     0.097    13.091    5.021     1016.131
6      3270     0.044    46.332    20.149    1745.389
7      9831     0.064    57.778    48.015    4564.360
8      29514    0.501    155.338   158.442   7755.513
9      88563    2.276    198.215   857.381   12081.491
10     265710   7.439    940.484   2693.862  15053.265
11     797151   27.578   1368.750  8771.571  15862.919
12     2391474  128.611  3279.699  nan       nan
}\data
\begin{document}
\begin{tikzpicture}[
    slanted blocks/.style={
        draw, fill=white, font=\tiny\ttfamily, rotate=45,
        anchor=south west, inner sep=2pt
    }]
    \begin{axis}[
        axis line style = {thick, gray},
        ymode = log,
        ymin = .2,
        xtick = {1,2,...,12},
        xlabel = {Depth},
        ylabel = {memory (MB)},
        legend style = {nodes={right, font=\scriptsize},
            at={(0.05,0.9)}, anchor=west},
        clip mode = individual,
        grid = major,
        ]
        \addplot table[x=depth, y=Sm]{\data};
        \addplot table[x=depth, y=Dm]{\data};
        \legend{Static memory, Dynamic memory}
        \pgfplotsinvokeforeach{0,...,11}{
            \node[slanted blocks] at ({#1+1},.2)
                {\pgfplotstablegetelem{#1}{Nodes}\of{\data}\pgfplotsretval};
        }
        \node[slanted blocks] at (0,.2) {\# of nodes};
        \node[starburst, fill=yellow, draw=red, thick, font=\tiny\ttfamily,
            inner sep=0pt, starburst point height=6pt]
            at (12, 2e4) {crash!};
    \end{axis}
\end{tikzpicture}
\end{document}

Anyway, I think that before starting coding the plot in LaTeX, you should get for example gnuplot, or matlab, and try to decide which graph you want, then code it.

Answer 2

Here's a way to do it, without spending too much time into refinement (e.g. the 10 s axis could be highlighted, or ranges could be set equal, to show the vast difference; put a legend etc.) :

Because you didn't know for my last answer how to give credit to my solution, here's just the starting code for you, so you can claim having made progress:

\documentclass[10pt,border=3mm,tikz]{standalone}
\usepackage{tikz}
\usetikzlibrary{datavisualization}
\begin{document}
\begin{tikzpicture}
    \datavisualization[scientific axes,
...
    data{
        x,      y
...
    };
 \end{tikzpicture}
\end{document}

Some unfinished indication, what you could do using some more properties of said package and placing two nodes and a line: