Oracle 10g Data Mining faq

What is the target market?
Oracle Data Mining addresses the needs of the data analyst as well as the application developer. ODM can be applied to a wide range of datasets (in terms of number of records and columns) and customer problems. It is especially suited for companies that have large data, are committed to the Oracle platform, and want to automate and operationalize their extraction of business intelligence. The initial end user is the data analyst and Java or PL/SQL application developer, although the end user of the application enhanced by data mining could be a customer service rep, marketing manager, customer, business manager, or just about any other imaginable user.

数据挖掘研究院

Since the results of data mining are in the database, these results are available to any other user or application. Oracle Data Mining helps leverage your investment dollars by making the new business information available to everyone. 数据挖掘研究院

Why would a business benefit from using Oracle Data Mining?
Data mining can sift through massive amounts of data and find hidden information — valuable information that can help you better understand your customers and anticipate their behavior. Oracle Data Mining software helps you build applications to uncover this hidden information about your customers. Armed with this information, you can build a close relationship with and understand your customers, which helps you to: 数据挖掘研究院

• Better retain customers and avoid churn
• Profile customers and understand their behavior
• Maintain and improve profit margins
• Reduce customer acquisition costs
• Target profitable customers with the right offer

• Predicting the quality of a manufactured part
• Finding associations between patients, drugs, and outcomes
• Identifying possible network intrusions

• Data Mining Embedded in Oracle Database
• Ability to Enhance Applications with Predictions and Insights
• Programming Interfaces
  
• Well-integrated with other Oracle products

数据挖掘研究院

1. Data Mining Embedded in Oracle 10g Database
   By being embedded in the Oracle 10g Database, Oracle Data Mining facilitates extracting business intelligence from large volumes of data for production applications. It eliminates off-loading data to external special-purpose analytic servers for data mining and the subsequent scoring of even larger volumes of data. The data, data preparation, data mining, and scoring all exist within the database, which greatly simplifies the data mining process.  In addition, overall application security is increased since data need never leave the secure database environment.

   Oracle Data Mining is available on all platforms supported by Oracle. This provides the widest range of platform support of any competitive data mining vendor. 

   数据挖掘研究院

   Oracle Data Mining can scale to the size of the problem by adding hardware or switching to more powerful platforms. Oracle Data Mining takes advantage of Oracle′s parallelism for faster computing by leveraging Oracle′s Real Application Clusters (RAC) technology.

   数据挖掘研究院

2. Ability to Enhance Applications with Predictions and Insights
   Oracle Data Mining enables companies to systemize the extraction and integration of new business intelligence within their operations. Application developers can use Oracle Data Mining′s Java-based interface to add data mining insights and predictions to enhance business applications, such as Customer Relationship Management (CRM), Enterprise Resource Management (ERP), Web portals, and wireless applications. If customers wish to migrate their data mining application to another platform, their investment is preserved. 

   Rather than having special departments of advanced data analysts who work on ad hoc data mining projects, the true value of data mining is realized when the new insights and predictions are integrated and "operationalized" into existing business applications.  数据挖掘研究院

   Telecommunications companies, for example, can use Oracle Data Mining to build churn applications that identify customers that are likely to churn before they leave for a competitor. Oracle Data Mining′s predictions can be used to anticipate customer behavior and proactively manage them in mutually beneficial 1:1 relationships.

   数据挖掘研究院

3. Programming Interfaces
   
   Application developers access Oracle Data Mining′s functionality through a Java-based or PL/SQL interface. Programmatic control of all data mining functions enables automation of data preparation, model-building, and model-scoring operations in production applications.
   
   Java Interface
   Java enables the development of platform independent applications and draws on SUN′s Java Community Process (JCP) for evolving and extending the language. Java-based applications can leverage the J2EE and J2SE platforms supported by a wide variety of vendors. Java is supported in the Oracle database which enables Java applications to run inside or outside the database. With the Oracle Application Server, developers can build web-based, distributed applications leveraging the J2EE platform within the Oracle product suite.
   数据挖掘研究院
   
   
   The Java API allows application programmers to control all aspects of the data mining process, from data preparation and model building to model testing and data scoring. The interface can be used at two levels. For the data mining expert, the Java interface exposes algorithm-specific settings for advanced users. For data mining novices, default values are provided for nearly all settings to minimize the specification.
   
   Programmatic control extends from data preparation and model building to on-demand scoring of single records and batch scoring of large data sets. Batch scores may be stored in relational tables for access by other business applications (e.g. call centers or marketing campaign systems) or called "on-demand" in interactive applications where new information is collected and provided to the predictive model, i.e. real-time scoring.
   
   Oracle Data Mining′s interface provides an early look at concepts and approaches being proposed for the Java Data Mining (JDM) standard. Ultimately, Oracle Data Mining will comply with the standard after it is published. JDM is an emerging data mining standard, following SUN′s Java Community Process as a Java Specification Request (JSR). JDM has participation from Oracle, Sun, IBM, SAS, SPSS, and many other companies that recognize the need for a Java- based standard for specifying and using data mining. JDM leverages several other data mining standards, including Object Management Group′s Common Warehouse Metadata (CWM), the Data Mining Group′s Predictive Model Markup Language (PMML), and International Standards Organization′s SQL/MM for Data Mining. 数据挖掘研究院
   
   PL/SQL Interface
   PL/SQL enables database developers to seamlessly integrate data mining functionality with their database applications. Oracle database developers and users are able to perform data mining operations using a familiar language and development methodology. Mining operations are presented as primitive capabilities, clearly separating transformations from model building and apply. PL/SQL introduces new capabilities not present in the Java API including native export/import of all supported models between ODM databases and advanced model evaluation using receiver operating characteristics (ROC).
   
   
   
4. Well-integrated with other Oracle Products
   
   By virtue of being in the Oracle 10g database, data preparation can be performed in Oracle 10g Warehouse Builder and deployed as part of an overall data mining application. Users can take advantage of key analytical and statistical capabilties in the Oracle 10g database to analyze data and results. 数据挖掘研究院
   
   Oracle 10g Data Mining is fully integrated with Oracle Universal Installer and RDBMS upgrade/downgrade process. This reduces the overhead associated with migrating applications between releases of non-Oracle data mining products.
   
   Oracle10g Data Mining is integrated with Oracle 11i Applications Suite, such as Oracle Marketing. 

What are some typical applications that could be enhanced by Oracle Data Mining?
Oracle Data Mining can automate the extraction and integration of new insight and predictions into a variety of business applications, including call centers, Web sites, campaign management systems, ATMs, enterprise resource management (ERM), and other operational and business planning applications. 数据挖掘研究院

数据挖掘实验室

• Support Vector Machine (SVM) algorithm supporting Regression and Classification mining functions
• Non-negative Matrix Factorization (NMF) algorithm supporting Feature Extraction
  
• Text Mining to support mining of structured and unstructued data
• BLAST algorithm supporting sequence matching and alignment searches
  
• PL/SQL interface
• Scoring Engine installation option
  数据挖掘研究院
  
  
• ROC computation
  
• 
  
  
Oracle Data Miner

Classification: Oracle Data Mining′s Classification algorithms can predict binary or multi-class outcomes. In binary problems, each record either will or will not exhibit the modeled behavior. For example, a model could be built to predict whether a customer will churn or remain loyal. These algorithms can also make predictions for multi-class problems where there are several possible outcomes. For example, a model could be built to predict which class of service will be preferred by each prospect. 数据挖掘实验室

Binary model example:
Q: Is this customer likely to become a high-profit customer?
A: Yes, with 85% probability 数据挖掘研究院

Multi-class model example:
Q: Which one of five customer segments is this customer most likely to fit into — Grow, Stable, Defect, Decline, or Insignificant?
A: Stable, with 55% probability

Algorithm options: Support Vector Machines (SVM), Naive Bayes, and Adaptive Bayes Networks (ABN) 数据挖掘实验室

Regression: Regression creates predictive models. The difference between regression and classification is that regression deals with numerical/continuous target attributes, whereas classification deals with discrete/categorical target attributes. In other words, if the target attribute contains continuous (floating-point) values, a regression technique is required. If the target attribute contains categorical (string or discrete integer) values, a classification technique is called for. 数据挖掘研究院

The most common form of regression is linear regression, in which a line that best fits the data is calculated, that is, the line that minimizes the average distance of all the points from the line. 数据挖掘研究院

This line becomes a predictive model when the value of the dependent variable is not known; its value is predicted by the point on the line that corresponds to the values of the independent variables for that record. Oracle Data Mining provides both linear and non-linear regression models.

Algorithm options: Support Vector Machines (SVM)

Association Rules: Association Rules detect "associated" or co-occurring events hidden in data. Association analysis, or unsupervised learning, is often used to find popular bundles (e.g. market basket analysis) of products that are related for customers, such as "milk" and "cereal" being associated with "bananas." Oracle Data Mining′s Association Rules can be used to identify co-occurring items or events in a variety of business problems, such as: 数据挖掘研究院

• Which manufactured parts and equipment settings are associated with failure events?
• Which patient and drug attributes are associated with which outcomes?
• Which items or products is this person most likely to buy or like?

Clustering: Oracle Data Mining provides two Clustering algorithms for the segmentation of individuals (cases) in a dataset. Both the enhanced version of K-Means and the proprietary O-Cluster algorithms produce clusters with descriptive rules for membership. Clusters are organized hierarchically with histograms provided for each attribute in a cluster. 数据挖掘研究院

Algorithm options: Enhanced K-Means and O-Cluster

数据挖掘研究院

数据挖掘研究院

Does Oracle Data Mining support PMML? 数据挖掘研究院
Oracle Data Mining supports the Data Mining Group′s Predictive Model Markup Language (PMML) for Naive Bayes and Association Rules models. Users can import and export models from these algorithms between Oracle database instances. Oracle Data Mining can import PMML from other vendors for these algorithms provided no vendor-proprietary extensions have been used.

数据挖掘研究院

Can ODM import PMML models from other vendors?
If other vendors generate standard core PMML for Naive Bayes and Association Rules, ODM should be able to import these models. PMML defines numerous optional capabilities which can make it difficult to ensure a given vendor′s PMML can be consumed. Some vendors use proprietary extensions to PMML which makes the models impossible to import.

What programming interfaces does Oracle Data Mining provide?
Oracle Data Mining provides two programming language interfaces: Java and PL/SQL. See APIs.
数据挖掘研究院