.png){kind=icon}
.png){kind=icon}
Return
Machine learning based success prediction for crowdsourcing software projects
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jss.2021.110965.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Competitive Crowdsourcing Software Development is an online software development paradigm, promises the innovative, cost effective and high quality solutions on time. However, the paradigm is still in infancy and does not address the key challenges such as low rate of submissions and high risk of project failure. A significant number of software projects fail to receive a satisfactory solution and end up wasting the time and efforts of stakeholders. Therefore, the success prediction of a new software project may help stakeholders in the project crowdsourcing decision, saving their time and efforts. To this end, this study proposes a novel approach based on machine learning to predict the success of a software project for crowdsourcing platforms in terms of whether the given project will reach its completion or otherwise. First, the textual description and important attributes of software projects from TopCoder is extracted. Next, the description is preprocessed using natural language processing technologies. Then, keywords are identified using a modified keyword ranking algorithm and each software project is awarded a ranking score. Every software project is modeled as a vector that is based on the extracted attributes, its identified keywords and ranking scores. Using these vectors with their associated solution status, a support vector machine classifier is trained to predict the success of a given software project. Different machine learning classifiers are applied and it turns out that support vector machine yields the highest performance on the given dataset. Finally, the proposed approach is evaluated with history data of real software projects. The results of hold-out validation suggest that the average precision, recall, and f-measure are up to 94.53%, 99.30% and 96.85%, respectively. (C) 2021 Elsevier Inc. All rights reserved.
Keywords:
Competitive crowdsourcing
Classification
Machine learning
Risk
Prediction
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.1
Papers:
5.4K
Citations:
8.4K