liveBook · Manning

Continuous hypothesis space

As explained in Grokking Bayes

A hypothesis space whose candidate values form a continuous numerical range rather than a finite collection.

A continuous hypothesis space contains candidate values across a numerical range, so nearby possibilities such as p = 0.2 and p = 0.200001 can differ infinitesimally. section 3.2

Instead of adding probabilities for finitely many hypotheses, use an integral to measure the area under a probability curve. The total area must equal 1. section 3.2

Keep these distinctions in mind:

  • A probability mass function assigns exact probabilities to discrete hypotheses. section 3.2
  • A probability density function describes relative likelihood across continuous hypotheses; integrating it gives the probability of a range. section 3.2

A probability between zero and one

A uniform continuous distribution over 0 to 1 treats every candidate value as equally likely, with its density normalized by integration. section 3.2

Use the equation view to inspect how integration replaces summation and normalizes continuous probability. section 3.2

How Grokking Bayes covers it

How deep it goes
Explains continuous ranges, integral normalization, and PDF meaning while contrasting them with discrete PMFs.
Examples
Uses probabilities 0.2 and 0.200001 and a uniform distribution from 0 to 1.
What you can do after
Evaluate/understand continuous beliefs by interpreting densities and integrating probability over value ranges.
How it connects
Connects probability distributions with integral calculus, probability density functions, probability mass functions, and uniform distributions.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Grokking Bayes. The section links point to the parts of the book it draws on.