Revision Questions: Classification Tree Method

Purpose

27 reflection questions for exam preparation. Use alongside sn_classification_tree.md for detailed explanations. Questions focus on understanding and application, not memorization.

Part 1: Introduction & Motivation

  1. Why was CTM developed? What specific problems with Category-Partition does it address?
  2. A colleague says “We don’t need CTM — our testers are experienced and know what to test.” How would you respond using evidence?
  3. Evaluate the strength of empirical evidence for CTM. What are its limitations?
  4. Why do the industrial results showing >90% test reduction not necessarily mean CTM is 90% more efficient?
  5. How does CTM relate to the broader context of combinatorial testing?

Part 2: Core Concepts

  1. A test object violates the “restful” property — it retains state between invocations. What problems does this cause for CTM?
  2. Given this specification, identify the test-relevant aspects and their classes: “A shipping calculator charges based on package weight (up to 50kg), destination (domestic/international), and delivery speed (standard 5-day, express 2-day, overnight). Packages over 50kg are rejected.”
  3. Explain the difference between composition and classification using a real-world example not from the lectures.
  4. Why must invalid cases be in a separate branch? Show what goes wrong with a concrete example if you put them inside each classification.
  5. Why are only classes test-selectable, and not compositions or classifications?
  6. What is the relationship between equivalence partitioning (EP) and CTM classes?

Part 3: The CTM Process

  1. You’re testing a mobile banking app. How does choosing “the transfer screen” vs “the entire app” as test object change your classification tree?
  2. Apply probing questions to the function search(database, query, filters, sortBy) and build a classification tree.
  3. Why does Grindal et al. recommend “Avoid” as the best constraint strategy? When might you choose “Replace” instead?
  4. Compare minimality and pairwise (2-wise) generation. When would you choose each?
  5. What does E[T] = 0.26 mean in practical terms? How would you improve it?
  6. Chen et al. proved two formal guarantees about tree restructuring. Why are these guarantees important?

Part 4: Examples & Practice

  1. Build a classification tree for an ATM withdrawal function. Include at least 4 aspects and model invalid cases.
  2. In the tax example, what would go wrong if you modeled Residency, Marital Status, and Gross Pay as a flat tree (all at root level) instead of nesting Status and Pay under the Resident class? How does hierarchy eliminate infeasible combinations?
  3. You have this abstract test specification: “Array: size > 1, unsorted, multiple occurrences, alphanumeric content. CountWhat: multiple characters.” Provide three different valid concrete test cases.
  4. In the password validation example, why can’t you use standard CTM combination generation? What special handling is needed?
  5. Identify the constraints (infeasible combinations) in the count() function tree. How would you handle them?

Part 5: Advanced Topics & Tools

  1. You have a web form with 4 tabs, each with different fields. Should you create one big tree or four separate trees? What are the trade-offs?
  2. Compare CTE XL and TESTONA. What capabilities does TESTONA add that make it more powerful for modern testing?
  3. A novice tester creates a flat tree with 5 classifications at the root level, each with 4 classes. What problems will they face? How would you coach them?
  4. Which of the five common pitfalls is most dangerous, and why?
  5. How does CTM integrate with equivalence partitioning, boundary value analysis, and combinatorial testing? Draw the connections.

Key Numbers to Memorize

Metric Value Source
Students preferring CTM 66% Yu et al. 2004
Study participants 162 (104 + 58) Yu et al. 2004
Training time in study 3 hours Yu et al. 2004
Ad hoc fault detection 55% of faulty programs Yu et al. 2004
Novices with infeasible cases 60% Yu et al. 2004
Subjectivity in tree structure 25% Yu et al. 2004
Test reduction (CTE XL) >90% (70,000 → 5,560) Lehmann 2000
Test suites evaluated (constraints) 3,854 Grindal et al. 2007
E[T] ad hoc tree 0.26 (74% waste) Chen et al. 2000
E[T] restructured tree 0.47 (53% waste) Chen et al. 2000
Best constraint strategy Avoid Grindal et al. 2007

Key Terms Glossary

Term Definition
Classification Tree Graphical model of the input domain with hierarchical aspects and classes
Test Object The unit under test — must be invocable, observable, and restful
Aspect / Classification A test-relevant property that influences behavior
Equivalence Class A set of input values expected to produce equivalent behavior
Composition “Has-a” node — children coexist simultaneously
Classification “Is-a” node — exactly one child applies
Class Leaf node — a selectable value in the combination table
Combination Table Matrix mapping classes to test specifications
E[T] Tree effectiveness = legitimate cases / potential cases
Dependency Rule Constraint marking infeasible class combinations
Probing Questions Systematic technique to discover test-relevant aspects
TSL Test Specification Language — text format used in Category-Partition
CTE XL Classification Tree Editor eXtended Logics — tool with constraint engine
TESTONA Modern CTM tool with SAT solver, oracles, and script generation
N-wise Generation strategy ensuring every n-tuple of classes appears
Minimality Generation strategy ensuring every class appears at least once

For detailed explanations, see study notes: sn_classification_tree.md

This site uses Just the Docs, a documentation theme for Jekyll.