td-outlier-detection

name	td-outlier-detection
description	Outlier detection and handling using TD_OutlierFit

Teradata Outlier Detection

Skill Name	Teradata Outlier Detection
Description	Outlier detection and handling using TD_OutlierFit
Category	Preprocessing Analytics
Function	TD_OutlierFit

Core Capabilities

• Complete analytical workflow from data exploration to model deployment
• Automated preprocessing including scaling, encoding, and train-test splitting
• Advanced TD_OutlierFit implementation with parameter optimization
• Comprehensive evaluation metrics and model validation
• Production-ready SQL generation with proper table management
• Error handling and data quality checks throughout the pipeline
• Business-focused interpretation of analytical results

Table Analysis Workflow

This skill automatically analyzes your provided table to generate optimized SQL workflows. Here's how it works:

1. Table Structure Analysis

• Column Detection: Automatically identifies all columns and their data types
• Data Type Classification: Distinguishes between numeric, categorical, and text columns
• Primary Key Identification: Detects unique identifier columns
• Missing Value Assessment: Analyzes data completeness

2. Feature Engineering Recommendations

• Numeric Features: Identifies columns suitable for scaling and normalization
• Categorical Features: Detects columns requiring encoding (one-hot, label encoding)
• Target Variable: Helps identify the dependent variable for modeling
• Feature Selection: Recommends relevant features based on data types

3. SQL Generation Process

• Dynamic Column Lists: Generates column lists based on your table structure
• Parameterized Queries: Creates flexible SQL templates using your table schema
• Table Name Integration: Replaces placeholders with your actual table names
• Database Context: Adapts to your database and schema naming conventions

How to Use This Skill

1. Provide Your Table Information:

   "Analyze table: database_name.table_name"
   or
   "Use table: my_data with target column: target_var"
   

2. The Skill Will:

   • Query your table structure using SHOW COLUMNS FROM table_name
   • Analyze data types and suggest appropriate preprocessing
   • Generate complete SQL workflow with your specific column names
   • Provide optimized parameters based on your data characteristics

Input Requirements

Data Requirements

• Source table: Teradata table with analytical data
• Target column: Dependent variable for preprocessing analysis
• Input data: Properly formatted Teradata table
• Column specifications: Appropriate data types for analysis

Technical Requirements

• Teradata Vantage with ClearScape Analytics enabled
• Database permissions: CREATE, DROP, SELECT on working database
• Function access: TD_OutlierFit

Output Formats

Generated Tables

• Preprocessed data tables with proper scaling and encoding
• Train/test split tables for model validation
• Processed results from TD_OutlierFit analysis
• Summary statistics and transformation parameters

SQL Scripts

• Complete workflow scripts ready for execution
• Parameterized queries for different datasets
• Table management with proper cleanup procedures

Preprocessing Use Cases Supported

1. Anomaly detection: Comprehensive analysis workflow
2. Data quality: Comprehensive analysis workflow
3. Outlier removal: Comprehensive analysis workflow

Best Practices Applied

• Data validation before analysis execution
• Proper feature scaling and categorical encoding
• Train-test splitting with stratification when appropriate
• Cross-validation for robust model evaluation
• Parameter optimization using systematic approaches
• Residual analysis and diagnostic checks
• Business interpretation of statistical results
• Documentation of methodology and assumptions

Example Usage

-- Example workflow for Teradata Outlier Detection
-- Replace 'your_table' with actual table name

-- 1. Data exploration and validation
SELECT COUNT(*),
       COUNT(DISTINCT your_id_column),
       AVG(your_target_column),
       STDDEV(your_target_column)
FROM your_database.your_table;

-- 2. Execute complete preprocessing workflow
-- (Detailed SQL provided by the skill)

Scripts Included

Core Analytics Scripts

• preprocessing.sql: Data preparation and feature engineering
• table_analysis.sql: Automatic table structure analysis
• complete_workflow_template.sql: End-to-end workflow template
• prediction.sql: TD_OutlierTransform execution
• evaluation.sql: Model validation and metrics calculation

Utility Scripts

• data_quality_checks.sql: Comprehensive data validation
• parameter_tuning.sql: Systematic parameter optimization
• diagnostic_queries.sql: Model diagnostics and interpretation

Limitations and Disclaimers

• Data quality: Results depend on input data quality and completeness
• Sample size: Minimum sample size requirements for reliable results
• Feature selection: Manual feature engineering may be required
• Computational resources: Large datasets may require optimization
• Business context: Statistical results require domain expertise for interpretation
• Model assumptions: Understand underlying mathematical assumptions

Quality Checks

Automated Validations

• Data completeness verification before analysis
• Statistical assumptions testing where applicable
• Model convergence monitoring during training
• Prediction quality assessment using validation data
• Performance metrics calculation and interpretation

Manual Review Points

• Feature selection appropriateness for business problem
• Model interpretation alignment with domain knowledge
• Results validation against business expectations
• Documentation completeness for reproducibility

Updates and Maintenance

• Version compatibility: Tested with latest Teradata Vantage releases
• Performance optimization: Regular query performance reviews
• Best practices: Updated based on analytics community feedback
• Documentation: Maintained with latest ClearScape Analytics features
• Examples: Updated with real-world use cases and scenarios

---

This skill provides production-ready preprocessing analytics using Teradata ClearScape Analytics TD_OutlierFit with comprehensive data science best practices.