Warwick, United Kingdom, August 15, 2010 --(PR.com
)-- Clustering will form part of the FastStats Modelling module, which currently includes a Profiling tool using Apteco’s PWE (Predictive Weight of Evidence) technique, a Decision Tree tool (using PWE or CHAID) and a Model Testing and Reporting tool. Currently in Beta testing, Apteco will be making this new functionality fully available for production use in the November 2010 FastStats software release.
Cluster analysis is about exploring and identifying natural groupings in a set of data points. James Alty, Managing Director at Apteco comments: “In FastStats our marketing and analytical users are interested in taking a selection of customers and automatically detecting groups with similar characteristics. Our new clustering tool will make it much easier for them to visualise their data and segment it into key target groups for marketing communication purposes. Once a set of clusters have been created, FastStats allows users to quickly apply the model across the whole database and allocate records to the appropriate cluster. The resulting data can be used directly for targeted selections. The power and immediate usability of the clustering analysis means we believe it will become a key tool within the FastStats Modelling module.”
FastStats Clustering uses the well-proven K-Means technique to allocate each point to its nearest cluster centre. However, the FastStats implementation introduces a number of powerful options including a multi-stage (Divisive or Agglomerative) clustering process. The user can also control the maximum number of iterations allowed, calculate the cluster proximity measure using Euclidean or City Block approaches, and finally define how the initial cluster centres are calculated (Frequency or Random).