Hands-On Unsupervised Learning with Python

Mode collapse

Given a probability distribution, the value that occurs most often (in the discrete case), or the one that corresponds to a maximum of the probability density function (in a continuous scenario), is called the mode. If we consider the latter case, a distribution whose PDF has a single maximum is called unimodal. When there are two local maxima, it's called bimodal, and so on (in general, when there are many modes, the distribution is simply called multimodal). Two examples are shown in the following screenshot:

Examples of unimodal (left) and bimodal (right) distributions

When working with complex datasets, we cannot easily estimate ...