Rich Techniques
The marketing analysis tools
provided by Discoverer are extensive and powerful. Recency Frequency Value
and Transaction Sequence analysis enables detailed understanding of customer
behaviour and their
transactional activity. Profiling Analysis with patented PWE Modelling and
Scoring enables a sophisticated understanding of customer characteristics.
Compare multiple
profiles!
Transaction and Basket Analysis
FastStats Discoverer includes the unique
basket and transaction analysis mechanism. Understanding the patterns of
transactions undertaken by your customers enables you to market to specific
purchase behaviours and devise more appropriate cross-sell and up-sell
offers. Insight into current transaction patterns can help you
predict the most likely next transaction for each customer and segment and
develop
marketing treatment on this basis.
Decision Tree Analysis
The Decision Tree option complements
the existing profiling tool. Profiling provides an easy to understand
uni-variate scoring mechanism, while the Decision Tree provides a sequenced
segmentation process which the user can control and resulting in descriptive
rules for each segment.
The Decision Tree module is so named because of the branching nature of the
statistics method used to identify segments of the database that fit certain
characteristics.
The Decision Tree initially compares the set of data to be
analysed with the base and then sequentially segments the data into branches
according to one of a number of statistical results. This splitting is to
identify groups of records that contain proportionally more or less of the
analysis data. FastStats Discoverer Decision Tree uses Apteco’s (patent pending)
Predictive Weight of Evidence score. Subsequent releases will include CHAID and
other split strategy algorithms.
The
Decision Tree module displays the results in a number of ways, both graphical
and tabular to assist the user visualise and interpret the results.
The Box Tree display shows a regular
branching diagram with labelling to identify the data used to split at that
branch and an indication of the value groupings used. The colour of each box
shows the proportion of successful records in each segment.
An innovative display called the
Organic Tree shows the same basic information in an easy to assimilate manner.
Like real trees, the Organic Tree display builds from the bottom up, the width
of each branch showing the number of records processed, the angle showing the
relative improvement and the colour showing the proportion of successful
records.
Marketing analysts now have new horticultural opportunities, perhaps
skilled analysts can even take up topiary as a hobby!
The
new ‘Organic Tree’ display provides an easy way for analysts and non technical
staff alike to get an appreciation of the segmentation achieved by the Decision
Tree.
The design and development of the
Decision Tree module has focussed on ensuring powerful statistics are available
within the easy and speedy style of FastStats Discoverer.
Decision Tree users
can interact with the tree building process by pausing or continuing
segmentation from particular nodes. Once they are happy with the segmentation
achieved, they can simply drag nodes from any of the tree displays or use a
gains table and lift chart to identify the most successful nodes.
This style of
operation fits perfectly with the ‘do anything, to anything’ style of the
Discoverer interface.
Click here for next page...
