A continuous random variable represents an uncertain quantity that can take values throughout a numerical range. A weather app might use one to describe how much rain will fall today, rather than only whether rain will occur. section 1.2
This gives a model more granularity: rainfall can be any non-negative amount, while a binary variable records only sunny or rainy. More detail can improve usefulness, but practical models still balance information against simplicity. section 1.3 section 1.2
The same weather question can be represented at different levels of detail:
- Binary variable: 0 means sunny and 1 means rainy. section 1.2
- Categorical variable: rainfall is rounded into a set of possible values. section 1.3
- Continuous variable: rainfall is modeled across all possible non-negative rates. section 1.2
From rainfall amount to rain probability
A detailed rainfall model can also answer the simpler question: the probability of zero rainfall represents sunny weather, while the combined probability of all positive rainfall rates represents rainy weather. section 1.2
