Introduction to Mining Model Content

Mining model content represents the patterns, rules, and relationships discovered by SQL Server Analysis Services (SSAS) data mining algorithms. This content is crucial for understanding the insights derived from your data, validating model accuracy, and making predictions.

Each mining model stores its discovered knowledge in a structured format, allowing you to query, visualize, and integrate it into your applications. Understanding the different types of content available for each algorithm is key to leveraging the full power of SSAS data mining.

Viewing Mining Model Content

You can view mining model content through several methods:

  • SQL Server Management Studio (SSMS): Provides a graphical interface to explore mining models, visualize their structure, and inspect their content.
  • Data Mining Extensions (DMX): Use DMX `SELECT` statements to query the mining model content directly, retrieving specific attributes and values.
  • XML for Analysis (XMLA): Programmatically access mining model content using XMLA protocols.
  • Client Applications: Integrate content into your reporting or business intelligence solutions.

The specific tables and views available for querying depend on the mining model type.

Types of Mining Model Content

The content of a mining model varies based on the algorithm used. Here are some common types:

Association Rules Models

  • Itemsets: Sets of items that frequently appear together.
  • Association Rules: Rules showing the relationship between itemsets (e.g., "If A, then B").
  • Support: Frequency of an itemset.
  • Confidence: Probability that if a transaction contains the antecedent, it also contains the consequent.

Clustering Models

  • Clusters: Groups of similar data points.
  • Cluster Characteristics: Attributes that define a cluster.
  • Cluster Profiles: Summary statistics for each cluster.

Classification and Regression Models

  • Decision Trees: Tree structures representing rules for prediction.
  • Linear Regression Equations: Mathematical formulas for predicting continuous values.
  • Predictors: Attributes that influence the prediction.

Sequence Clustering and Mining Models

  • Sequences: Ordered patterns of events.
  • Transitions: Probabilities of moving from one state to another.

Using Mining Model Content

Mining model content can be used for various purposes:

  • Understanding Data: Gain insights into relationships and patterns.
  • Model Evaluation: Assess the quality and accuracy of your models.
  • Generating Predictions: Use discovered rules or models to predict future outcomes.
  • Business Decision Making: Inform marketing strategies, product recommendations, and operational improvements.
  • Drill-through: Navigate from summary-level model content to the underlying source data.

Examples

Querying Association Rules Content with DMX

To retrieve association rules with at least 80% confidence:

SELECT RULE_SET, RULE_PROBABILITY, RULE_SUPPORT
FROM [YourMiningModelName].RULES
WHERE RULE_PROBABILITY > 0.8

Querying Cluster Characteristics with DMX

To retrieve characteristics for a specific cluster:

SELECT ATTRIBUTE_NAME, ATTRIBUTE_VALUE, ORGANIZATION
FROM [YourClusteringModelName].CLUSTER_CHARACTERISTICS
WHERE CLUSTER_NUMBER = 3

Explore the specific documentation for each data mining algorithm to understand the full scope of its content and available querying options.