Abstract

Due to immense growth in digital technology and high-end intelligence algorithms, the software industry has grown immensely in a multi-fold manner over the years. This immense growth involves thousands of software engineers and developers to contribute their skills in coding in the process of product development, delivery, deployment, and services. More often, due to the wrong estimation of the software efforts, the development companies may have deep holes in their pockets. Traditional techniques like the COCOMO model and others are working in a sequential manner to assess the software effort estimation; this was good in an earlier way of software development. Due to the evolution of cloud computing, the software development process has become deeply distributed. Hence, an intelligent way is required to assess the software effort estimation process, which involves the deep learning model and more. So, this survey paper focuses primarily on evaluating the earlier works in software effort estimation and trying to find the major gaps to propose a better model to achieve good accuracy. This research paper proposes a strategy for finding the software effort estimation by using support vector machine model with a hybrid model powered by an artificial neural network (ANN) and long-short-term memory (LSTM) to catalyze the decision tree model to produce good accuracy on software effort estimation.