linear regression vs logistic regression

Linear regression

• prediction is continuous. For example the prediction is in a scale of 0-100

• a simple algorithm that models a linear relationship between inputs and a continuous numerical output variable.

• use case:
  • stock price prediction
  • predicting housing prices
  • predicting customer lifetime value

• pros:
  • explainable method
  • interpretable results by its output coefficients
  • faster to train than other machine learning models

• cons:
  • assumes linearity between inputs and outputs
  • sensitive to outliers
  • can underfit with small, high-dimensional data

Logistic regression

• prediction is discrete. For example the prediction is the student can pass or fail the exam.

• a simple algorithm that models a linear relationship between inputs and categorical output(0/1)

• use case:
  • credit risk score prediction
  • customer churn prediction

• pros:
  • interpretable and explainable method
  • less prone to overfitting when using regulation
  • applicable for multi-class predictions

• cons:
  • assuming linearity between input and output
  • can overfit with small, high dimensional data