TOEFLify: An Adaptive English Learning Application

01 / Process

How the work unfolded

TOEFLify was developed using the Prototyping Model methodology, an iterative approach that ensures the application meets user needs through continuous feedback and refinement.

The development process involved several key stages:

• Needs Analysis: We began by identifying core user requirements, designing the initial application specifications, and planning the database structure.
• Prototype Design: This stage focused on creating wireframes and designing the intuitive user interface (UI).
• Prototyping Implementation: The core features were built through coding. This included implementing a machine learning-based CEFR level prediction algorithm, building the UI, and integrating generative AI to create adaptive practice questions.
• Iteration and Refinement: The application was refined based on user feedback and evaluations. A System Usability Scale (SUS) test was conducted with 10 respondents, resulting in an average score of 77.5. This score indicates the application is usable, but with some areas for improvement.
• Data Processing Methodology: Data was managed using the CRISP-DM (Cross-Industry Standard Process for Data Mining) approach. This involved:
  • Modeling: We employed several machine learning algorithms, including Logistic Regression, Support Vector Machine (SVM), Random Forest, and Naive Bayes.
  • Evaluation: The performance of each model was evaluated using metrics such as accuracy, precision, and F1-score. The Logistic Regression model showed the highest accuracy at 81.18%.