An evolutionary morphological approach for software development cost estimation

Area of Science:

• Computer Science
• Artificial Intelligence
• Software Engineering

Background:

• Software Development Cost Estimation (SDCE) is a critical but challenging problem.
• Traditional methods often struggle with the complexity and non-linear nature of SDCE.
• Existing artificial neuron models face limitations in gradient estimation for morphological operators.

Purpose of the Study:

• To propose an evolutionary morphological approach for SDCE.
• To introduce a novel hybrid artificial neuron, the dilation-erosion perceptron (DEP).
• To develop an evolutionary learning process, DEP(MGA), to overcome gradient estimation issues.

Main Methods:

• Developed a dilation-erosion perceptron (DEP) based on mathematical morphology (MM) and complete lattice theory (CLT).
• Implemented a modified genetic algorithm (MGA) for an evolutionary learning process (DEP(MGA)) to train the DEP model.
• Conducted experimental analysis on five complex SDCE problems using three standard performance metrics.

Main Results:

• The DEP model demonstrated good performance in solving complex software development cost estimation problems.
• The evolutionary learning process (DEP(MGA)) effectively addressed the non-differentiability of morphological operators.
• Experimental results validated the efficacy of the proposed hybrid approach.

Conclusions:

• The evolutionary morphological approach, specifically the DEP model trained with MGA, offers a viable solution for SDCE.
• This method provides an effective way to handle the complexities inherent in software cost estimation.
• The DEP model shows potential for improving the accuracy and reliability of SDCE.