Common practice in Machine Learning often implicitly assumes a stationary distribution, meaning that the distribution of a particular feature does not change over time. In practice, however, this ass
