Toad/SegmentationTree

Segmentation Tree

This is my implementation of what Vic described in WikiTargeting/DynamicSegmentation.

The tree is binary, each internal node represent a segmentation choice. For exemple a DistanceNode? would discriminate data according to a distance threshold. At the moment the internal node type used are

• DistanceNode

• VelocityNode

• LateralVelocityNode

• AccelerationNode

• TimeSinceVelocityChangeNode

• WallNode

• WallReverseNode

  .

The leaves of the tree contains the observations that fit in the segmentation obtained by following the path from the root to the leaf. When a leaf contains enough observations it is split. The factors used for splitting a leaf are

• keeping approximately the same amount of data in each branch. It picks the median value for each segmentation axis, and corrects it to the closest standardised value. The standardised value allow the tree to be saved while maintaining a small file size (less than 600ko for a 35 rounds game even with long rounds).
• the information gain of all possible internal node type.

The tree is rebuild at the beginning of each round. When there is too much data to rebuild the tree completely it is just partially rebuild.

The tree can be saved at the end of a game, in fact just a symbolic representation of the internal nodes, at the moment this is disabled I haven't tested if it is valuable.

---

Comments & Questions