Software cost estimation predication using a convolutional neural network and particle swarm optimization algorithm

Area of Science:

• Computer Science
• Software Engineering

Background:

• Software cost estimation is crucial for project planning and resource allocation.
• Existing estimation methods suffer from inaccuracy and instability, necessitating advanced techniques.

Purpose of the Study:

• To develop and evaluate a novel model for software cost estimation using hybrid deep learning and machine learning techniques.
• To enhance prediction accuracy and reduce manual effort in the software cost estimation process.

Main Methods:

• A hybrid model integrating Convolutional Neural Networks (CNN) for feature extraction and Particle Swarm Optimization (PSO) for hyperparameter tuning was developed.
• The model was trained and validated on 13 benchmark datasets, employing time series forecasting principles.
• Performance was assessed using Mean Absolute Error (MAE), Mean Square Error (MSE), Mean Magnitude Relative Error (MMRE), Root Mean Square Error (RMSE), Median Magnitude Relative Error (MdMRE), and Prediction Accuracy (PRED).

Main Results:

• The proposed CNN-PSO model demonstrated superior performance compared to existing methods across all 13 datasets and six evaluation metrics.
• The model achieved higher prediction accuracy and improved robustness and generalization capabilities.

Conclusions:

• The hybrid CNN-PSO model offers a promising and effective solution for accurate software cost estimation.
• This approach significantly reduces manual effort and enhances the reliability of software cost predictions.