Educational usability Analysis of WEKA and Orange for Learning Data Mining: A Task-Based Comparative Study with Students Feedback

Technical Summary: Educational Usability Analysis of WEKA and Orange for Learning Data Mining

Problem Statement
Data mining tools are essential for academic instruction and practical analysis, yet the selection of appropriate software for introductory courses remains a challenge. While tools like WEKA and Orange are widely used in educational institutions due to their open-source nature and user-friendly interfaces, previous comparative studies have predominantly focused on technical metrics such as algorithmic efficiency, predictive accuracy, and computational performance. There is a lack of research evaluating these tools specifically from the perspective of educational usability for beginner learners. This study addresses the gap by investigating which tool—WEKA or Orange—provides a more intuitive and educationally effective environment for students learning fundamental data mining concepts, including preprocessing, classification, and visualization.

Methodology
The research employed a task-based comparative study involving a modified version of the Iris Flower Dataset. To simulate real-world data quality issues, the dataset was intentionally altered to include missing values and outliers. The experimental environment utilized WEKA (Version 3.8) and Orange (Version 3.35) with the Naive Bayes classification algorithm.

The study followed a four-stage experimental workflow:

1. Data Ingestion and Exploration: Assessing the ease of loading datasets and understanding initial data structures.
2. Data Preprocessing: Evaluating the accessibility of cleaning mechanisms, such as handling missing values and outliers.
3. Model Construction: Analyzing the process of configuring and executing classification algorithms.
4. Evaluation and Visualization: Reviewing how results, confusion matrices, and graphical representations are presented.

The evaluation criteria were grounded in educational theory rather than raw performance metrics. Specifically, the study assessed:

• Interface and Navigation Transparency: How clearly the software communicates the next steps in the mining process.
• Cognitive Load: Based on Sweller's Cognitive Load Theory, the study measured whether the interface minimized unnecessary mental effort related to software mechanics, allowing students to focus on concepts.
• Visual Feedback and Workflow Clarity: The ability of the tool to visually illustrate data progression from raw input to final output.

To validate these technical observations, a questionnaire-based feedback session was conducted with eight undergraduate students who had prior exposure to both tools. The students were divided into two groups based on their learning speed (fast learners and those requiring more time) to ensure a diverse perspective on usability.

Key Results
The comparative analysis revealed distinct differences in how the two tools support beginner learners:

• Interface and Workflow: WEKA utilizes a card-based, menu-driven interface (e.g., "Preprocess," "Classify" tabs) that presents raw statistical data immediately. While detailed, this text-heavy layout was found to create an early cognitive load, requiring students to interpret abstract statistics before understanding the dataset's physical shape. In contrast, Orange employs a visual workflow canvas where users drag and drop widgets (e.g., "File," "Preprocess") onto a blank space. This approach allows students to construct their workspace step-by-step, explicitly separating the data file from the actions performed on it.
• Preprocessing: In WEKA, applying filters requires navigating complex, nested text directories (e.g., Preprocess -> filters -> Unsupervised -> attributes). Orange simplifies this by using a checklist-style dialog box within a connected widget, making the sequence of data modification explicitly clear via visual data lines.
• Modeling and Evaluation: WEKA outputs classification results as a static, text-based stream containing detailed mathematical data (probabilities, standard deviations, confusion matrices). While professional, this format was described as overwhelming for beginners who struggle to parse "walls of text." Orange presents results in a clean, sortable tabular grid and visualizes the entire testing framework (Train/Test splits, Cross-validation) as a block diagram, breaking the machine learning lifecycle into clear, independent steps.
• Visualization: Orange offers superior interactivity, allowing students to view multiple graph types simultaneously and dynamically adjust axes. It also enables the direct connection of correlation tables to scatter plots, visually mapping coefficients next to data points. WEKA's visualization is functional but less interactive and requires navigating separate windows for different views.

Student Feedback and Ratings
The student feedback (n=8) quantitatively supported the qualitative observations. On a 5-point scale, Orange consistently outperformed WEKA across all educational usability criteria:

• Beginner Friendliness: Orange (4.75) vs. WEKA (3.00).
• Workflow Clarity: Orange (4.63) vs. WEKA (2.88).
• Understanding Data Mining Process: Orange (4.50) vs. WEKA (3.38).
• Visualization: Orange (4.38) vs. WEKA (3.38).

In terms of preference, 7 out of 8 students selected Orange as "Easier to Learn," "Better for Classroom Learning," and providing "Better Understanding of Data Mining." All 8 students preferred Orange for visualization and preprocessing support.

Significance and Claims
The paper concludes that while both WEKA and Orange successfully execute fundamental data mining tasks, they serve different educational roles. The study claims that Orange is more suitable for introductory data mining education due to its visual workflow, interactive interface, and ability to reduce cognitive load for beginners. The modular, drag-and-drop nature of Orange helps students visualize the data mining process and understand the sequence of operations without getting bogged down by complex menu hierarchies or dense text outputs.

Conversely, the study acknowledges that WEKA remains a reliable tool for learning traditional data mining techniques and is valuable for advanced students who require detailed, text-based statistical outputs. However, for the specific goal of teaching introductory concepts to novice learners, the study suggests that Orange's interface transparency and workflow clarity make it the superior choice for educators aiming to optimize the learning experience. The findings provide guidance for teachers and institutions in selecting software that aligns with the cognitive needs of beginner students.