Logistic regression
A very popular use of GLM is via logistic regression. This is the type of regression to use when your target variable is a binary value, that is, it can only take on two values. These two binary values usually take the form:
- Something occurred (binary value = 1)
- Something did not occur (binary value = 0)
These two values typically vary according to the type of industry. Here are some examples of binary responses in different industries:
- Marketing: Did the customer respond to a specific offer or not?
- Health care: Was a specific drug effective in treating a condition or was it not?
- Financial: Did a specific trading strategy lead to a profit or not?
- Web tracking: Did a customer abandon a shopping cart at the checkout or not?
Virtually every industry will have problems which can be formulated in terms of binary outcomes. In fact, sometimes standard linear regression problems are often recast, so instead of predicting the specific values for the target value, the problem will be reformulated...
