Classification Tree Method

What Is the Classification Tree Method? #

The Classification Tree Method (CTM) is a visual test design technique developed at the German software testing institute (DaimlerChrysler). It provides a structured way to decompose the input domain of a test object into a tree of classifications and classes, then generate test cases by selecting combinations from the tree.

Key Concepts #

• Test object: The root node — the system, function, or feature being tested
• Classification: A test-relevant aspect or dimension (like a parameter category)
• Class: A specific value or partition within a classification
• Combination table: A matrix below the tree showing which classes combine into test cases

Classification Tree Structure #

The tree decomposes “Online Payment” into three classifications (Payment Method, Amount, Currency), each with their own classes.

Building a Classification Tree: Step by Step #

Step 1: Identify the test object and its test-relevant aspects.

Step 2: For each aspect, create a classification node.

Step 3: Under each classification, list the possible classes (values/partitions).

Step 4: Add refinements — classes can have sub-classifications if needed.

Step 5: Build the combination table to select test cases.

The Combination Table #

Below the tree, create a table where:

• Columns = test cases
• Rows = classes (leaf nodes)
• Marks (X or dot) show which class is selected for each test case

Each test case picks exactly one class from each classification.

CTM vs. Equivalence Partitioning #

CTM builds on EP but adds:

• Visual structure — the tree makes the decomposition clear
• Hierarchical refinement — classes can be further subdivided
• Explicit combination selection — the table shows exactly which combinations you’re testing

When to Use CTM #

• Complex input domains with multiple related parameters
• When you need a visual overview of the test space
• When stakeholders need to review test coverage visually
• As a complement to pairwise testing for selecting meaningful combinations

Advanced Classification Trees #

Hierarchical Refinement #

Classes can have sub-classifications, creating deeper trees:

This creates a richer decomposition where Business users are further classified into SMB and Enterprise, and Social Login into Google and GitHub.

Combining CTM with Pairwise #

For large trees, exhaustive combination of all classes produces too many test cases. Apply pairwise selection to the combination table:

1. Build the classification tree
2. List all leaf-node classes as parameters
3. Feed into PICT or generate pairwise combinations manually
4. Fill the combination table with pairwise-selected test cases

This gives you the visual benefits of CTM with the efficiency of pairwise testing.

Tool Support: Classification Tree Editor (CTE) #

CTE XL is a dedicated tool for creating classification trees:

• Visual tree editor with drag-and-drop
• Automatic combination generation (exhaustive, pairwise, etc.)
• Export to test management tools
• Available as free academic version

Real-World Example: E-Commerce Search #

Test Object: Product Search

Classifications:
├── Search Query
│   ├── Single word
│   ├── Multiple words
│   ├── With special characters
│   └── Empty
├── Filters
│   ├── No filters
│   ├── Category filter
│   ├── Price range filter
│   └── Multiple filters
├── Sort Order
│   ├── Relevance
│   ├── Price (low to high)
│   ├── Price (high to low)
│   └── Newest
└── Pagination
    ├── First page
    ├── Middle page
    └── Last page

With 4 x 4 x 4 x 3 = 192 exhaustive combinations, pairwise on this tree produces approximately 16-20 test cases.

Exercise: Build a Classification Tree #

Scenario: You’re testing a file export feature with these aspects:

• File format: PDF, CSV, Excel, JSON
• Data range: Last 7 days, Last 30 days, Custom range, All data
• Content: Summary only, Detailed, With charts
• Delivery: Download, Email, Cloud storage

Tasks:

1. Draw the classification tree
2. Create a combination table with at least 6 test cases
3. Ensure every class appears in at least one test case (minimum coverage)

File Export
├── Format: PDF, CSV, Excel, JSON
├── Range: 7 days, 30 days, Custom, All
├── Content: Summary, Detailed, With Charts
└── Delivery: Download, Email, Cloud

Pro Tips #

• Start with the test object clearly defined. A vague root leads to a vague tree. “Login” is better than “User Management.”
• Keep classifications independent. If two classifications are highly dependent, consider merging them or adding constraints.
• Use CTM for communication. The visual tree is excellent for review meetings — stakeholders can quickly spot missing aspects.
• Refine iteratively. Start with a simple tree, then add sub-classifications as you learn more about the system.
• Mark invalid combinations. In the combination table, mark impossible combinations (like JSON + Charts) to prevent generating useless test cases.