Bernoulli Distribution - Python and Machine Learning for Integrated Circuits - - An Online Book - |
||||||||
Python and Machine Learning for Integrated Circuits http://www.globalsino.com/ICs/ | ||||||||
Chapter/Index: Introduction | A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | Z | Appendix | ||||||||
================================================================================= The Bernoulli distribution is a fundamental probability distribution in statistics and probability theory. It models a random experiment with two possible outcomes: success (usually denoted as 1) and failure (usually denoted as 0). The distribution is named after Swiss mathematician Jacob Bernoulli (refer to page3887). The Bernoulli distribution is characterized by a single parameter, which is the probability of success, often denoted as "p." The probability mass function (PMF) of the Bernoulli distribution is given by: Where:
Based on Equation 3865a, we can have, Then, comparing with the Exponential Family equation below: we have: h(x) = 1 θ = log(p/(1-p)) x = T(x) A(θ) = log(1-p) = -log(1-1/(1+exp(-θ))) The probability distribution of the dependent variable in logistic regression follows: 1) A Bernoulli distribution for binary logistic regression. 2) A categorical distribution for multinomial logistic regression. ============================================
|
||||||||
================================================================================= | ||||||||
|
||||||||