Scholarlink

<p>All approaches, and life cycles in software development need Software Effort Estimation (SEE). The accuracy of estimations is critical for effective project timeline and resource management. The current SEE methodology comparison review is incomplete due to the diverse sample methodologies and performance criteria defined by different scholars. This study aims to fill this gap by offering a precise effort estimation model based on deep learning techniques principles. The aim is to reduce development costs and time. The data undergoes parsing to eliminate non-informative portions and to capture important and higher-order statistical characteristics. An innovative hybrid optimization algorithm, Modified Chaotic Enriched Jaya with Moth Flame Optimization (MCEJ), was developed for the purpose of extreme feature selection with high precision in estimating SEE framework, hence incorporating accuracy into the model. The estimation is done through Multilayer Long Short-Term Memory (M-LSTM) network. The implementation was done using Python, achieving a 0.2825 MSE with Dataset 1 (China) and 0.2285 with Dataset 2 (Maxwel).</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Novel Hybrid Deep Learning Framework with Metaheuristic Optimization for Accurate Software Effort Estimation

  • 期刊论文

  • 发表日期：

    2025年10月16日

查看全文
  • Mahesh Badana,
  • B. Kiran Kumar,
  • M. Rekha Sundari,
  • P. Ramya,
  • G. Sanjiv Rao

摘要

All approaches, and life cycles in software development need Software Effort Estimation (SEE). The accuracy of estimations is critical for effective project timeline and resource management. The current SEE methodology comparison review is incomplete due to the diverse sample methodologies and performance criteria defined by different scholars. This study aims to fill this gap by offering a precise effort estimation model based on deep learning techniques principles. The aim is to reduce development costs and time. The data undergoes parsing to eliminate non-informative portions and to capture important and higher-order statistical characteristics. An innovative hybrid optimization algorithm, Modified Chaotic Enriched Jaya with Moth Flame Optimization (MCEJ), was developed for the purpose of extreme feature selection with high precision in estimating SEE framework, hence incorporating accuracy into the model. The estimation is done through Multilayer Long Short-Term Memory (M-LSTM) network. The implementation was done using Python, achieving a 0.2825 MSE with Dataset 1 (China) and 0.2285 with Dataset 2 (Maxwel).