DeepLearningDNNLinearCombinedRegressorconstruct a combined neural network and linear regressor
Calling SequenceParametersOptionsDescription
<Text-field style="Heading 2" layout="Heading 2" bookmark="usage">Calling Sequence</Text-field>
DNNLinearCombinedRegressor(lfc,dfc,opts)
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk0">Parameters</Text-field>lfc-list of linear feature columnsdfc-list of deep neural network feature columnsopts-(optional) one or more keyword options described below
<Text-field style="Heading 2" layout="Heading 2" bookmark="bkmrk1">Options</Text-field>
hidden_units : list of nonnegative integers
Specifies the number of hidden nodes in the neural network at each level.
optimizer : function or Optimizer object, or list of two functions or Optimizer objects
Specifies the optimizer to use to train the model. When a list of two optimizers [opt1,opt2] is provided, the first is taken to be the linear optimizer and the second the neural network optimizer. The default uses the FTRL optimizer for linear optimization and the Adagrad optimizer for neural network optimization.
<Text-field style="Heading 2" layout="Heading 2" bookmark="info">Description</Text-field>
The DNNLinearCombinedRegressor(lfc,dfc,opts) command creates a combined linear and neural network regressor for the feature columns specified in lfc and dfc.
Linear regression is performed on the features in lfc and neural network regression is performed on the features in dfc.
This function is part of the DeepLearning package, so it can be used in the short form DNNLinearCombinedRegressor(..) only after executing the command with(DeepLearning). However, it can always be accessed through the long form of the command by using DeepLearning[DNNLinearCombinedRegressor](..).
See AlsoDeepLearning OverviewDeepLearning[DNNLinearCombinedClassifier]DeepLearning[DNNRegressor]DeepLearning[LinearRegressor]