Oracle Generalized Linear Models (GLM)

(11g only) Generalized linear models relax the restrictive assumptions made by linear models. These include, for example, the assumptions that the target variable has a normal distribution, and that the effect of the predictors on the target variable is linear in nature. A generalized linear model is suitable for predictions where the distribution of the target is likely to have a non-normal distribution, such as a multinomial or a Poisson distribution. Similarly, a generalized linear model is useful in cases where the relationship, or link, between the predictors and the target is likely to be non-linear.

For more information, see the Oracle Data Mining Application Developer’s Guide and Oracle Data Mining Concepts.