Generalized Linear Models

What is a Generalized Linear Model?

Generalized linear models are an expansion of linear regressions, which allow different output distribution functions to describe the variance of observations from the predicted values. For example, if the output distribution is normal, then it is a classic linear regression, where a constant change in input variables leads to a constant change in the output value. In the case of generalized linear models, there is link function defines the relationship between inputs and output. A standard example of a link function is an exponential function of the response’s density function, enabling the variance of each observation to be a function of its predicted value.


Why are Generalized Linear Models Important?

Generalized linear models can support a wider variety of problems than linear regression. They allow more flexibility in the types of functions used for linking inputs and outputs and in the kinds of variables that can be predicted, including real numbers, integers, positive real or integer, Boolean, or sets (categories)..


How Enables Organizations to Use Generalized Linear Models makes it easy to apply generalized linear models to address domain-specific applications of AI to deliver business value today. The C3 AI® Suite is a complete, end-to-end platform for designing, developing, deploying, and operating enterprise AI applications at industrial scale. supports generalized linear models to analyze data within an ML pipeline using either the low-code C3 AI ML Studio development environment or the no-code C3 AI Ex Machina tool. Within Ex Machina, for example, a business analyst can graphically link a data set to an analytics model to train a generalized linear model to determine the optimal predictive parameters, and then apply that model to new incoming data to predict values of results, all without writing a line of code.