Executing Decision Rule Algorithms in WEKA

• Course: Databases and Data Mining
• Practical Session
• Instructor: Jamolbek Mattiev

Learning Objectives

• • Understand rule-based classifiers
• • Execute rule algorithms in WEKA
• • Interpret generated rules
• • Compare algorithm performance

What Are Decision Rules?

• • IF–THEN classification rules
• • Easy to interpret
• • Alternative to decision trees
• • Human-readable knowledge extraction

Common Rule Algorithms in WEKA

• • OneR (One Rule)
• • JRip (RIPPER)
• • PART (Partial Decision Trees)
• • DecisionTable

Step 1: Open WEKA Explorer

• • Launch WEKA
• • Click Explorer
• • Load dataset (.arff)
• • Go to Classify tab

Step 2: Select Rule Algorithm

• • Click Choose
• • Select → rules
• • Choose OneR / JRip / PART

Step 3: Set Evaluation Method

• • Percentage split (70%)
• or
• • 10-fold cross-validation

Step 4: Execute Algorithm

• • Click Start
• • Observe generated rules
• • Record accuracy
• • Analyze confusion matrix

Understanding Output

• • Generated IF–THEN rules
• • Number of rules
• • Accuracy (%)
• • Kappa statistic

Example Rule (JRip)

• IF Outlook = Sunny AND Humidity = High
• THEN Play = No
• ELSE Play = Yes

Comparison Criteria

• • Accuracy
• • Number of rules
• • Model interpretability
• • Computational cost

Advantages of Rule-Based Models

• • Transparent
• • Easy to explain
• • Suitable for knowledge discovery

Limitations

• • Can overfit
• • Sensitive to noise
• • May generate many rules

Discussion Questions

• • Which algorithm performs best?
• • Which produces simpler rules?
• • When prefer rules over trees?

Summary

• • Rule algorithms produce IF–THEN models
• • Easy to implement in WEKA
• • Performance varies by dataset
• • Compare using validation methods

Example Performance Comparison of Rule Algorithms