Introduction to the ML life cycle
The ML life cycle is a continuous process that a data science project follows. It contains four major stages, starting with data collection and preparation, model training, model evaluation, and finally model inferencing and monitoring. The ML process is a continuous one, where the cycle iterates between improving the data and constantly improving the model's performance; or, rather, keeping it from degrading over time:
Figure 9.1 – ML life cycle
The previous diagram presents the continuous process of ML life cycle management, from data preparation to model development, and then from training to model deployment and monitoring. When model performance degrades due to either a change in the training data or the model code or changes in model parameters, the cyclic process starts all over again.
Processes for data collection and preparation, cleansing, and consolidation, as well as techniques for training various...