A Systematic Literature Review of Machine Learning Techniques for Software Effort Estimation Models

Abstract

Estimating software projects is a challenging but necessary process in software development. Predicting the effort needed to build software is an essential part of the project life cycle. This paper examines a variety of machine learning algorithms for estimating effort. There has been a significant increase in research on effort estimation with machine learning approaches during the last two decades, with the objective of improving estimation accuracy. To forecast effort, the estimation techniques such as expert judgment, COCOMO, analogy based, putnam model, and machine learning are used. The algorithmic models' low accuracy and unreliable architecture resulted in substantial software project risks. As a result, it is essential to predict the cost of project on an annual basis and compare it to alternative methods. However, the effort prediction using machine learning is still limited because a single technique cannot be treated as best. This paper's main goal is to present a review of several machine learning approaches for predicting effort.