This document discusses descriptive versus mechanistic modelling in the pharmaceutical industry. Descriptive models are used to describe data without understanding the underlying mechanism, while mechanistic models attempt to understand the data-generating process. Examples are provided of using non-parametric techniques like kernel regression for descriptive modelling and differential equation models for mechanistic modelling of tumor growth curves. The conclusion emphasizes that mechanistic models going beyond simple data fitting to incorporate scientific knowledge can help justify models and improve understanding of biological processes.