Logistic Regression

Introduction: Logistic Regression

Logistic regression is a statistical method used to model the probability of a binary outcome, such as yes/no, success/failure, or pass/fail. Instead of predicting continuous values, it estimates the likelihood that an event occurs, making it a key tool in decision-making and quality improvement.

Background

Developed as an extension of regression analysis, logistic regression became widely used in the 20th century for analysing categorical data. Unlike linear regression, which predicts values on a continuous scale, logistic regression applies the logit function to transform probabilities into a linear relationship. This makes it suitable for modelling outcomes constrained between 0 and 1.

Key Elements / Features

• Dependent variable: Binary (two categories, e.g., defect/no defect).
• Independent variables: Can be continuous (temperature, time) or categorical (machine type, operator).
• Logit function: Converts probabilities into log-odds for analysis.
• Interpretation: Coefficients indicate how each input variable influences the odds of the outcome.

Applications / Examples

• Healthcare: Predicting whether a patient will recover based on treatment and risk factors.
• Business: Estimating the probability of customer churn.
• Quality control: Assessing the likelihood that a product passes or fails inspection.
  Example: A factory applies logistic regression to estimate the probability that a machine produces a defect, based on operating speed and temperature.

Relevance / Impact

Logistic regression is widely applied because it accommodates both attribute and continuous predictors. It is simple to interpret yet powerful enough to form the foundation of many predictive analytics and machine learning models. In Lean Six Sigma, it is useful for linking process inputs to binary quality outcomes, supporting data-driven decision-making.

See also

• Regression Analysis
• Attribute Data
• Chi-Square Test
• Qualitative Data