For those who’ve been in the information science space for any period of time, you’ve most definitely heard this buzz term.
The machine learning life cycle.
It sounds fancy, but that is what it really boils right down to:
- Machine learning is an energetic and dynamic process — it doesn’t have a strict starting or end
- Once a model is trained and deployed, it is going to most definitely have to be retrained as time goes on, thus restarting the cycle.
- There are steps inside the cycle, nevertheless, that have to be followed of their proper order and executed rigorously
Once you Google the ML life cycle, each source will probably offer you a rather different variety of steps and their names.
Nevertheless, you’ll notice that for essentially the most part, the cycle accommodates: problem definition, data collection and preprocessing, feature engineering, model selection and training, model evaluation, deployment, and monitoring.
