|
Classic
Overview | Classic
Uniqueness | Classic
Key features | Classic
Userbase | Classic
System Requirements
Classic Overview
Classic is Corporate Intellect’s decision tree induction
algorithm. Classic can be used to discover classification
and regression patterns from vast data stores.
A typical data set provided as input to Classic consists
of a number of attributes, one of which is the target
attribute, that is, the attribute that needs to be predicted
based on the other attributes, referred to as independent
attributes, within the data. The target attribute can
either be categorical in nature, for example, insurance
policy lapse or renewal, or continuous, for example, price
of a commodity.
Given such input data, Classic learns a model from the
data in the form of a decision tree. An example decision
tree is shown below.
The tree above is part of a decision tree discovered for
an insurance company that wanted to identify the segments
of their customer base that were more likely to lapse
their motor insurance policy. This would allow the company
to build specific products to service these segments and
improve their retention rate. Current levels of acquisition
costs require a customer to remain loyal to the customer
for 6 years before they become profitable for the company.
Current churn rates are resulting in a fall in profits
within the motor insurance industry. The extract of the
tree shown above suggests that people over the age of
34 with smaller cars are likely to Renew as opposed to
Lapse their policy. The confidence in this result is 90%.
The algorithm uses information theoretic and statistical
metrics to choose the decision attributes at each node
within the tree, progressively partitioning the input
data into more homogenous data sets with regard to the
target attribute. Local models can then be built within
these data partitions (at the leaf nodes of the tree)
to predict the target attribute.
back to top
back to
products |
|
|
Coral
Summary
