Logistic regression

A regression for a yes-or-no outcome that models the log odds as linear in the variables, so predicted probabilities stay between 0 and 1. Estimated by maximum likelihood rather than least squares.

Where you will meet it

Logistic regression appears in 2 learning outcomes on the CFA syllabus. A few of them:

• Set out which investment questions multiple linear regression can answer and the steps involved in running one
  CFA Level II · Quantitative Methods

  “Frame the question, choose the dependent and independent variables, then ask whether the dependent variable is continuous; if it is a yes-or-no outcome, logistic regression fits better than ordinary least squares.”

• Build a logistic regression model and explain what its output means
  CFA Level II · Quantitative Methods

  “Logistic regression instead models the log odds: ln[p ÷ (1 − p)] = b0 + b1X1 + … + bkXk, estimated by maximum likelihood.”

See also

Log odds

Practise it

• Today's practice question One question, with every wrong option explained.
• Free Level I diagnostic test Twelve questions that show which topics to start with.
• Free finance calculators Check your arithmetic on the exam's core formulas.

All terms beginning with L · Glossary