this is for holding javascript data
Paul St-Aubin edited Experimental Results.tex
almost 10 years ago
Commit id: 5764779185474076e814af893dc9e2cecc06c298
deletions | additions
diff --git a/Experimental Results.tex b/Experimental Results.tex
index 572114f..d2893de 100644
--- a/Experimental Results.tex
+++ b/Experimental Results.tex
...
\subsection{Site Profiles and Clustering}
Kmeans gives the best clustering, with 5 clusters providing a good balance between size and distribution.
Table~\ref{tab:clustering_group_sizes} lists the cluster group frequency for each clustering attempt. One group, cluster 1, consistently stands out, alone. This site is the only on of the sample that contains a slip lane, which allows right-turning vehicles to avoid the roundabout merging zone entirely. This cluster is therefore rejected. A second group, cluster 2, is also clustered consistently. These three sites correspond to limited-access highway ramp roundabouts (that function as interchange junctions) and are characterised by no land use at all, single lanes, moderatly high approach speeds, and low intersection density. Curiously, they are never grouped together with roundabouts situated \textit{on} regional highways which are managed by the same transportation agency, built to the same standards and roughly share similar land usage.
\begin{table}[h]
\caption{Clustering Attempts}
\label{tab:clustering_group_sizes}
\begin{tabular}{llllll}
\hline
kmeans & Freq. (4) & Freq. (5) & Freq.(6) & Freq.(7) & Freq.(8) \\ \hline
1 & 1 & 1 & 1 & 1 & 1 \\
2 & 3 & 3 & 3 & 3 & 2 \\
3 & 6 & 22 & 17 & 12 & 3 \\
4 & 31 & 10 & 10 & 4 & 4 \\
5 & - & 5 & 5 & 5 & 5 \\
6 & - & - & 5 & 4 & 4 \\
7 & - & - & - & 12 & 12 \\
8 & - & - & - & - & 10
\end{tabular}
\end{table}
\begin{verbatim}