A multiple criteria decision analysis based approach to remove uncertainty in SMP models

Area of Science:

• Software Engineering
• Computer Science

Background:

• Software maintainability is critical for managing development lifecycle costs and organizational assets.
• Neglecting maintainability assessment before software release can lead to significant post-deployment expenses and reputational harm.
• Existing prediction models often struggle with heterogeneous datasets and new programming paradigms, posing challenges for small and medium-sized organizations.

Purpose of the Study:

• To develop a robust software maintainability prediction model applicable to heterogeneous datasets and various programming paradigms.
• To address the limitations of current models that hinder maintainability assessment in diverse software development environments.

Main Methods:

• Utilized a Genetic Algorithm (GA) to optimize a Random Forest technique for software maintainability prediction.
• Employed the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to select the optimal prediction model.
• Tested the model's efficacy on heterogeneous datasets encompassing different programming paradigms.

Main Results:

• The Genetic Algorithm optimized Random Forest model demonstrated high accuracy in predicting software maintainability.
• The proposed approach proved effective across diverse programming paradigms, overcoming limitations of previous methods.
• The study confirmed the GA's suitability for enhancing maintainability prediction in varied software development contexts.

Conclusions:

• The GA-optimized Random Forest model is optimal for predicting software maintainability, especially across heterogeneous datasets and multiple programming paradigms.
• This approach offers a viable solution for organizations, particularly small and medium-sized ones, to conduct accurate maintainability assessments.
• Prioritizing and effectively predicting software maintainability is essential for successful software development and deployment.