To examine the simulations, a number of visualisations were developed:

\begin{frame}{Quantisation}

% {{{

To aid compression, the floating point data is first converted to integer values.

\begin{figure}

  \centering

  \includegraphics[width=75mm]{min-images/quantisation.png}

  \caption{Quantising point data can be seen as snapping points to a grid.}

\end{figure}

\end{frame}

% }}}

  \item Researchers will be visually examining the simulations, how much error is noticeable?

  \item Determine an appropriate level of quantisation, tradeoff between compression ratio and discarded data. \\ \

\begin{frame}{Analysis of Quantisation Results}

% {{{

\begin{itemize}

  \item The datasets do \textbf{not} have a statistically significant effect \\

  {\footnotesize Which is \textbf{good}, all datasets were similary affected by quantisation.

  Friedman Test \textbf{p-value: $>$0.1}}

  % \item The data is normal \\

  % {\footnotesize Shapiro-Wilks Test \textbf{p-value: $<$0.01}}

  \item Quantisation has a statistically significant effect \\

  {\footnotesize ANOVA \textbf{p-value: $<$2.2e-16}}

  \item 8 and 10 bit quantisation are not statistically significantly different

  from each other \\

  {\footnotesize Ball-and-stick: \textbf{p-value of 0.7135299}} \\

  {\footnotesize Metaballs: \textbf{p-value of 0.9762546}}

\end{itemize}

\end{frame}

% }}}

\begin{frame}{Quantisation Results}

% {{{

\begin{figure}

  \centering

  \includegraphics[width=100mm]{min-images/bm-boxplot.png}

  \caption{Box-and-whisker plot showing the ratings for each of the quantisation levels and visualisations.}

\end{figure}

\end{frame}

% }}}