Contents

Title and Copyright Information

Send Us Your Comments

Preface

Intended Audience

Where to Find More Information

Documentation Accessibility

1 Introduction

1.1 ODM Requirements and Constraints

2 ODM Java Programming

2.1 Compiling and Executing ODM Programs

2.2 Using ODM to Perform Mining Tasks

2.2.1 Prepare Input Data

2.2.2 Build a Model

2.2.3 Find and Use the Most Important Attributes

2.2.4 Test the Model

2.2.5 Compute Lift

2.2.6 Apply the Model to New Data

3 ODM Java API Basic Usage

3.1 Connecting to the Data Mining Server

3.2 Describing the Mining Data

3.2.1 Creating LocationAccessData

3.2.2 Creating NonTransactionalDataSpecification

3.2.3 Creating TransactionalDataSpecification

3.3 MiningFunctionSettings Object

3.3.1 Creating Algorithm Settings

3.3.2 Creating Classification Function Settings

3.3.3 Validate and Store Mining Function Settings

3.4 MiningTask Object

3.5 Build a Mining Model

3.6 MiningModel Object

3.7 Testing a Model

3.7.1 Describe the Test Dataset

3.7.2 Test the Model

3.7.3 Get the Test Results

3.8 Lift Computation

3.8.1 Specify Positive Target Value

3.8.2 Compute Lift

3.8.3 Get the Lift Results

3.9 Scoring Data Using a Model

3.9.1 Describing Apply Input and Output Datasets

3.9.2 Specify the Format of the Apply Output

3.9.3 Apply the Model

3.9.4 Real-Time Scoring

3.10 Use of CostMatrix

3.11 Use of PriorProbabilities

3.12 Data Preparation

3.12.1 Automated Binning and Normalization

3.12.2 External Binning

3.12.3 Embedded Binning

3.13 Text Mining

3.14 Summary of Java Sample Programs

4 DBMS_DATA_MINING

4.1 Development Methodology

4.2 Mining Models, Function, and Algorithm Settings

4.2.1 Mining Model

4.2.2 Mining Function

4.2.3 Mining Algorithm

4.2.4 Settings Table

4.2.4.1 Prior Probabilities Table

4.2.4.2 Cost Matrix Table

4.3 Mining Operations and Results

4.3.1 Build Results

4.3.2 Apply Results

4.3.3 Test Results for Classification Models

4.3.4 Test Results for Regression Models

4.3.4.1 Root Mean Square Error

4.3.4.2 Mean Absolute Error

4.4 Mining Data

4.4.1 Wide Data Support

4.4.1.1 Clinical Data -- Dimension Table

4.4.1.2 Gene Expression Data -- Fact Table

4.4.2 Attribute Types

4.4.3 Target Attribute

4.4.4 Data Transformations

4.5 Performance Considerations

4.6 Rules and Limitations for DBMS_DATA_MINING

4.7 Summary of Data Types, Constants, Exceptions, and User Views

4.8 Summary of DBMS_DATA_MINING Subprograms

4.9 Model Export and Import

4.9.1 Limitations

4.9.2 Prerequisites

4.9.3 Choose the Right Utility

4.9.4 Temp Tables

5 ODM PL/SQL Sample Programs

5.1 Overview of ODM PL/SQL Sample Programs

5.2 Summary of ODM PL/SQL Sample Programs

6 Sequence Matching and Annotation (BLAST)

6.2 Using ODM BLAST

6.2.1 Using BLASTN_MATCH to Search DNA Sequences

6.2.1.1 Searching for Good Matches in DNA Sequences

6.2.1.2 Searching DNA Sequences Published After a Certain Date

6.2.2 Using BLASTP_MATCH to Search Protein Sequences

6.2.2.1 Searching for Good Matches in Protein Sequences

6.2.3 Using BLASTN_ALIGN to Search and Align DNA Sequences

6.2.3.1 Searching and Aligning for Good Matches in DNA Sequences

6.2.4 Output of the Table Function

6.2.5 Sample Data for BLAST

Summary of BLAST Table Functions

BLASTN_MATCH Table Function

BLASTP_MATCH Table Function

TBLAST_MATCH Table Function

BLASTN_ALIGN Table Function

BLASTP_ALIGN Table Function

TBLAST_ALIGN Table Function

7 Text Mining

A Binning

A.1 Use of Automated Binning

B ODM Tips and Techniques

B.1 Clustering Models

B.1.1 Attributes for Clustering

B.1.2 Binning Data for k-Means Models

B.1.3 Binning Data for O-Cluster Models

B.2.1 Build Quality and Performance

B.2.2 Data Preparation

B.2.3 Numeric Predictor Handling

B.2.4 Categorical Predictor Handling

B.2.5 Regression Target Handling

B.2.6 SVM Algorithm Settings

B.2.7 Complexity Factor (C)

B.2.8 Epsilon -- Regression Only

B.2.9 Kernel Cache -- Gaussian Kernels Only

B.2.10 Tolerance

Index