Of late, Agile methodologies have been taking root in data science boosting complex collaborations between data scientists and other developers. Agile can be easily ported over to Machine Learning and Artificial Intelligence domains due to its feedback-heavy, iterative nature and given that incessant improvement is an innate part of AI. Such methodologies are characterized by fast feedback loops and short development sprints. Agile projects, in distinction to old-school waterfall approaches, involve error correction and cyclical stakeholder input and primarily focuses on short term goals rather than the long-term view. AI researchers should think of research as an iterative, evolving process to remain receptive and adaptive as per Agile’s basic tenets. To ensure that projects do not grind to a halt, maintaining a buffer of solutions for implementation is a priority as data scientists work on multiple projects, each taking months to complete. The iterative nature of Agile well captures experimentation as a core part of AI and ML projects. Agile maximizes value throughout the development process.

Read More at: http://www.dataversity.net/case-agile-machine-learning/