Spatial Modeling of IPTV Potential A Case Study: Massillon Cable TV 2006 Location Intelligence Conference Professor Paul Rappoport, Temple University Robert.

Spatial Modeling of IPTV PotentialSpatial Modeling of IPTV Potential

A Case Study: Massillon Cable TVA Case Study: Massillon Cable TV

2006 Location Intelligence Conference

Professor Paul Rappoport, Temple UniversityRobert Gessner, President, Massillon Cable TV

Dr. Amy Liu, Marketing Systems GroupKevin Babyak, Marketing Systems Group

April 3-5, 2006San Francisco, California

OutlineOutline

• The ProblemThe Problem• The ApproachThe Approach• Case StudyCase Study• Results & ImplicationsResults & Implications

The ProblemThe Problem

How can a local cable provider measure the How can a local cable provider measure the competitive threat posed by a telephone competitive threat posed by a telephone competitor?competitor?Are all service areas equally at risk?Are all service areas equally at risk?

What customer segments are at risk?What customer segments are at risk?

How can spatial information of a market provide How can spatial information of a market provide competitive insight?competitive insight?Can advertising be used to effectively challenge Can advertising be used to effectively challenge competitor’s claims of Internet speed?competitor’s claims of Internet speed?

Where could IPTV be providedWhere could IPTV be provided

The ProblemThe Problem

The “claim” is that a telephone competitor The “claim” is that a telephone competitor can provide high speed Internet access as part can provide high speed Internet access as part of a package of services. However, current of a package of services. However, current technology is limited by distance – not all technology is limited by distance – not all households can receive high speed access. households can receive high speed access. Thus this claim could be disputed. Thus this claim could be disputed.

The ApproachThe Approach Nearest Neighbor Hierarchical ClusteringNearest Neighbor Hierarchical Clustering

Identify groups of households that are spatially close where close is based on 2 criteria:

1. Threshold distance – only points that are closer than the threshold distance are selected for clustering

2. A minimum number of households are required to form a cluster

These clusters can then be used to produce a hierarchy of clusters, where higher order clusters satisfy the above two criteria.

Cluster become entities for subsequent analyses.

Hierarchical ClustersHierarchical Clusters

First order clusters (the smaller circles) can be combined to form higher order clusters (the red ovals) and so forth until an entire market area is evaluated.

ApproachApproach

In this application, clusters provide a proxy for the presence of remote terminals or other outside plant that could be used to deliver high speed data or video services.

Cluster attributes include the number of homes passed, average income, penetration rates for DBS, broadband, average spending on video and local and long distance telephone.

Clusters can then be used to segment a market by degree of risk or contestability.

Massillon Cable TVMassillon Cable TV

Cable franchise area is defined in this analysis by block groups


The cable franchise has areas of very low to very high levels of income.


There are 5 central offices that coincide with the cable area. High speed Internet requires that a central office be enabled for providing DSL.


This map displays the distribution of households by ZIP+4. The majority of the 32,000 households are clustered in the City of Massillon


This map displays the results of the clustering as well as the location of remote terminals. Remote terminals can be used by a telephone company to extend the reach of DSL


The effective reach of current DSL technology is 3 KM. The circles display the reach of DSL


If selected remote terminals become DSL enabled, households in the higher income areas could receive high speed Internet access and other IPTV services from the telephone company.

Results & ImplicationsResults & Implications

Spatial clustering reduces the complexity of the problem Clusters represent entities for analyzing competitive activity Clusters can be evaluated by usage, spending and

demographic characteristics

Spatial clustering identifies areas that are contestable For large systems, this minimizes the type of competitive

response Clusters provide an efficient scaling for targeted marketing

Results & ImplicationsResults & Implications

For Massillon Cable TV, the analysis For Massillon Cable TV, the analysis uncovers areas that are contestableuncovers areas that are contestable

For Massillon Cable TV, these areas For Massillon Cable TV, these areas correspond to high income locations. correspond to high income locations. Broadband and video services are strongly Broadband and video services are strongly correlated with income.correlated with income.

70% of homes passed are potentially at risk70% of homes passed are potentially at risk

Contact InformationContact Information

Paul Rappoport [email protected]

Robert Gessner [email protected]

Amy Liu [email protected]

Kevin Babyak [email protected]

CitationCitation

The citation for the clustering algorithm is:

Ned Levine, CrimeStat III: A Spatial Statistics Program for the Analysis of Crime Incident Locations. Ned Levine and Associates, Houston, TX., and the National Institute of Justice, Washington, D.C. November 2004.

Spatial Modeling of IPTV Potential A Case Study: Massillon Cable TV 2006 Location Intelligence Conference Professor Paul Rappoport, Temple University Robert.

Documents