Breaking down regression models

While there are several regression model types available in the machine learning eco-system, there are two primary regression models groups: linear and logistic, both of which have rich implementations in ML.NET.

ML.NET provides the following linear regression trainers:

• FastTreeRegressionTrainer
• FastTreeTweedieTrainer
• FastForestRegressionTrainer
• GamRegressionTrainer

• LbfgsPoissonRegressionTrainer
• LightGbmRegressionTrainer
• OlsTrainer
• OnlineGradientDescentTrainer
• SdcaRegressionTrainer

The employee attrition application we will be creating later in this chapter utilizes the linear regression SDCA trainer.

In addition, ML.NET provides the following binary logistic regression trainers:

• LbfgsLogisticRegressionBinaryTrainer
• SdcaLogisticRegressionBinaryTrainer
• SdcaNonCalibratedBinaryTrainer
• SymbolicSgdLogisticRegressionBinaryTrainer

For the file classification application, we will be utilizing the SDCALogisticRegressionBinaryTrainer model.