Logit Function

Home Glossary Item Logit Function
« Back to Glossary Index

The logit function is a core component in statistics and machine learning, particularly in logistic regression models. It is the inverse of the sigmoid function and it mathematically expresses the logarithm of the odds ratio. The term “logit” comes from “log of odds” and it transforms a probability value ranging between 0 and 1 into a log-odds value on the real number line.


The main use of the logit function in machine learning is to convert a continuous input into a probability value that can be used for binary classification tasks. For example in logistic regression, the predictors’ linear combination is passed through the logit function to produce a probability. This output probability can then be used to classify whether an instance belongs to a particular class. Based on certain symptoms, a machine learning model can predict whether a patient has a specific disease.

The beauty of the logit function extends beyond its application in logistic regression. While predicting probabilities is essential in numerous practical situations, real-world events often do not follow a purely linear pattern. The logit function’s S-shaped curve allows better accommodation of complex relationships between the predictors and the predicted event. It does so by giving the outputs a probabilistic interpretation, thereby enhancing the model’s predictive capabilities for binary outcomes in various fields ranging from healthcare to finance.

« Back to Glossary Index