Classic Summary
Classic is the latest addition to the Corporate Intellect’s innovative suite of data mining algorithms aimed at classification and regression data mining tasks.
Classic belongs to the next generation of the decision tree induction family of algorithms. Classic leverages the simplicity of the decision boundaries defined by a decision tree induction algorithm along with local models to generate more accurate and complex decision surfaces.
Typical example application to which Classic can been applied include:
• Policy Lapse Modelling
• Claims Fraud
• Property Price Appraisal
• Money Laundering
• Churn Analysis
• Manufacturing Yield Enhancement
• Target Mailing
Classic Uniqueness | Key Features | Userbase | System Requirements
Classic Uniqueness
Fast, Robust and Scalable
Classic has been developed with speed, robustness and scalability as central to its design allowing it to run on data ranging from a few records to millions of records.
Generation of PMML output
The Predictive Modelling Markup Language (PMML) is a standard, XML representation for the knowledge discovered using data mining. Classic can produce the resulting decision tree in PMML enabling the easy exchange for knowledge produced by Classic and other data mining vendor applications and scoring engines.
Ability to assign models to the leaf nodes resulting in more complex models
Most decision tree induction algorithms generate a decision tree through the use of information theoretic and statistical measures, progressively partitioning the data provided as input to the algorithm, until further splitting of the data actually degrades the performance of the resulting knowledge on unseen data due to overfitting. Once the tree has been built, a new data element can be assigned to one of the leaf nodes based on the values of the independent attributes.
A unique feature of Classic is its ability to combine the simplicity of the decision tree univariate splits with other modelling paradigms assigned locally to the tree’s leaf nodes, creating more complex local decision surfaces and producing more accurate models from data.
back to top
Classic Key Features
Handling Missing Values
Classic uses surrogate predicates within each decision node to deal with missing data. Surrogate predicates are used when the data record being scored has a missing value for the split attribute of the node. The surrogate mimics the split effected by the primary split predicate of the node.
Binary and Multiple Splits
Classic provides the user with the power to adjust the level of bushiness of the resulting tree by adjusting the maximum number of branches emerging from each node.
Handle Continuous and Categorical Data
Classic automatically decides on the optimal binning of continuous attributes to maximise their contribution to the resulting gain in information by using it as the principal split or surrogate predicate.
Generation of PMML output
The Predictive Modelling Markup Language (PMML) is a standard, XML representation for the knowledge discovered using data mining. Classic can produce the resulting decision tree in PMML enabling the easy exchange for knowledge produced by Classic and other data mining vendor applications and scoring engines.
Fast, Robust and Scalable
Classic has been developed with speed, robustness and scalability as central to its design allowing it to run on data ranging from a few records to millions of records.
back to top
Classic Userbase
Classic provides a valuable data-mining algorithm for the following vertical markets:
• Telecommunications
• Health Care
• Banking Finance
• Insurance
• Re-Insurance
• Manufacturing
• Retail
• Consumer Packaged Goods
• Market Research
• Public Sector
• Academia
back to top
Classic System Requirements
Classic is written in Java and is therefore available for most hardware and software platforms. The recommended system requirements for installing and running Classic are:Hardware Pentium-compatible processor or higher and a monitor with 1024 x 768 resolution or higher (support for 65,536 colors is recommended). A CD-ROM drive for installation is also required.
Operating system Windows 98, Windows 2000, or Windows NT 4.0 with Service Pack 6 or higher. Solaris 2.6 is also supported.
Min free disk space 5MB is required for Classic.
Min RAM 128MB or more of RAM
back to top