Make machine learning projects work

Data scientists or machine learning practitioners often have the desire to try out algorithms and models that they perceive as sexier or more fashionable. This temptation is hard to resist and can sometimes be detrimental to a the success of a project, for such pursuit can lead to the failure of not completing the project on time and within budget. I believe one of the keys to successful ML project is having a structured and measured approach whereby project decisions — for example whether to implement a new deep neural network architecture — are made based on quantitative evaluation. My favourite resource on this topic comes from Andrew Ng in his “Machine learning yearning” book and his “Structuring machine learning projects” course on Coursera.

Both are fairly non-technical but offer useful techniques and practices to manage ML projects. Below are a few examples of what you’ll learn from the course:

• Understand how to diagnose errors in a machine learning system, and
• Be able to prioritize the most promising directions for reducing error
• Understand complex ML settings, such as mismatched training/test sets, and comparing to and/or surpassing human-level performance
• Know how to apply end-to-end learning, transfer learning, and multi-task learning