23DATA DISTRIBUTION SHIFTS
What are the main types of data distribution shifts we may encounter after model deployment?
Data distribution shifts are one of the most common problems when putting machine learning and AI models into production. In short, they refer to the differences between the distribution of data on which a model was trained and the distribution of data it encounters in the real world. Often, these changes can lead to significant drops in model performance because the model’s predictions are no longer accurate.
There are several types of distribution shifts, some of which are more problematic than others. The most common are covariate ...
Get Machine Learning Q and AI now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
