Tracking of Hardware Development Schedule based on Software Effort Estimation

Accurately predicting the time required for tasks in the development process can effectively manage resources and costs, which is crucial in project management. In 1960, Farr [3] and Nelson [6] proposed the concept of software effort estimation. Early research focused on building standardized estimation models to estimate the number of hours worked to complete tasks through statistical regression analysis or expert rules of thumb. Later, machine learning and deep learning were used to train models to estimate working hours to replace traditional estimation methods. This study proposes that software effort estimation can be extended to the hardware development industry. We use machine learning and deep learning to estimate the time required for tasks in the hardware development process and then accurately manage the product development time. This research uses semantic analysis to extract the keywords of the problems in the development process through NLP and use them as features for afterward analysis. We compare the accuracy, MMRE, and PRED(25) of the four models of machine learning's decision tree, random forest, XGBoost, and deep learning's RNN model in estimating the time required for tasks. The experimental results show that the decision tree has higher accuracy than the other three models. This study proves that the software effort estimation technique can be applied to task tracking in the hardware development process.