Batch Processing

Gradient boosted trees classification follows the general workflow described in Training and Prediction > Regression > Usage Model and Training and Prediction > Classification and Regression > Gradient Boosted Trees.

Training

For the description of the input and output, refer to Training and Prediction > Regression > Usage Model. In addition to parameters of the gradient boosted trees described in Training and Prediction > Classification and Regression > Gradient Boosted Trees > Batch Processing, the gradient boosted trees classification training algorithm has the following parameters:

Parameter

Default Value

Description

algorithmFPType

float

The floating-point type that the algorithm uses for intermediate computations. Can be float or double.

method

defaultDense

The computation method used by the gradient boosted trees regression. The only training method supported so far is the default dense method.

nClasses

Not applicable.

The number of classes. A required parameter.

loss

crossEntropy

Loss function type.

Prediction

For the description of the input and output, refer to Usage Model: Training and Prediction.

In addition to the parameters of the classifier, the gradient boosted trees classifier has the following parameters at the prediction stage:

Parameter

Default Value

Description

algorithmFPType

float

The floating-point type that the algorithm uses for intermediate computations. Can be float or double.

method

defaultDense

The computation method used by the gradient boosted trees regression. The only training method supported so far is the default dense method.

nClasses

Not applicable.

The number of classes. A required parameter.

numIterations

0

An integer parameter that indicates how many trained iterations of the model should be used in prediction. The default value 0 denotes no limit. All the trained trees should be used.

For more complete information about compiler optimizations, see our Optimization Notice.
Select sticky button color: 
Orange (only for download buttons)