WCB Cost Model Virtual Workshop 2012 - Clustering

Please provide comments to the issue below as part of the 2012 WCB cost model virtual workshop for inclusion in the record. Comments are moderated for conformity to the workshop’s guidelines.

Background

Hybrid Cost Proxy Model: In developing the HCPM, the Commission concluded that a clustering algorithm should group customer locations into serving areas in an efficient manner to minimize costs while maintaining a specified level of network performance quality. Efficient clustering of customers leads to a least-cost and most-efficient network design that is consistent with forward-looking modeling principles.

The HCPM limits clusters by engineering constraints. It attempts to group customers so that they are no further away than allowed by network design, and so that no more customers are attached to a digital loop carrier remote terminal (DLC) than is permitted by network design. Initially, the model groups all customer locations into one cluster, then divides the cluster if one or more of the engineering constraints is violated. The model then uses several optimization routines that reassign certain customers to different clusters in order to lower the cost of constructing distribution areas.

CQBAT: The CQBAT model largely uses the same clustering approach as the HCPM, but uses road-based routing to determine the maximum size of the clusters. Thus, clusters defined by CQBAT are likely smaller but more realistic estimates of cluster size; by using road segments in clustering, CQBAT avoids the problem of having the length of some loops along roads exceed maximum loop length.

Questions for Comment

1. The approach taken by CQBAT appears to be appropriate. Is there any reason to deviate from this approach? Are there any incremental improvements that should be made to CQBAT's clustering approach?