this is for holding javascript data
Lucas Fidon edited subsection_Estimated_MI_metric_for__.tex
almost 8 years ago
Commit id: 2be5b2569ac8d62122ea6c0486b99c277003539c
deletions | additions
diff --git a/subsection_Estimated_MI_metric_for__.tex b/subsection_Estimated_MI_metric_for__.tex
index a3e9d61..7816a2e 100644
--- a/subsection_Estimated_MI_metric_for__.tex
+++ b/subsection_Estimated_MI_metric_for__.tex
...
In the case of trajectories clustering we have used two different metrics based on the previous MI-based metric.
\begin{enumerate}
\item MI-metric based on the probability distribution of
player's player position during a few minutes
\item MI-metric based on the probability distribution of
player's player acceleration during a few minutes
\end{enumerate}
However we don't know the probability distributions of our problem,
hence so we have to approximate those distributions using the trajectories.
Thus Therefore, we will use estimated MI metric in both cases.
Position
parameters have parameter has been choose because it is the most intuitive and
user-friendly. user-friendly available parameter.
Regarding the acceleration
parameters, they have parameter, is has a Physics interpretation. Indeed if we have the Fundamental principle of the dynamics in mind we can relate it to a kind of strength which acts on the players. So in this case the players belonging to the same cluster using this acceleration MI based distance may be the players with a "common strength" acting on them.