Base class for parameters of the gradient boosted trees classification training algorithm.
More...
◆ getFeaturesPerNode()
long getFeaturesPerNode |
( |
| ) |
|
Returns number of features tried as possible splits by the gradient boosted trees classification training algorithm If 0 then sqrt(p) is used, where p is the total number of features.
- Returns
- Number of features
◆ getLambda()
Returns lambda parameter of the training procedure (L2 regularization on weights lambda) Range: [0, inf). Default is 1
- Returns
- lambda value
◆ getMaxBins()
Returns maximal number of discrete bins to bucket continuous features. Used with 'inexact' split finding method only. Default is 256. Increasing the number results in higher computation costs
- Returns
- Maximal number of discrete bins to bucket continuous features
◆ getMaxIterations()
long getMaxIterations |
( |
| ) |
|
Returns maximal number of iterations parameter of the gradient boosted trees training algorithm
- Returns
- Number of iterations
◆ getMaxTreeDepth()
Returns maximal tree depth. Default is 0 (unlimited)
- Returns
- Maximal tree depth
◆ getMinBinSize()
Returns minimal number of observations in a bin. Default is 5 Used with 'inexact' split finding method only.
- Returns
- Minimal number of observations in a bin
◆ getMinObservationsInLeafNode()
long getMinObservationsInLeafNode |
( |
| ) |
|
Returns minimal number of samples per node. Default is 1
- Returns
- Minimal number of samples
◆ getMinSplitLoss()
double getMinSplitLoss |
( |
| ) |
|
Returns loss regularization parameter of the training procedure. Min loss reduction required to make a further partition on a leaf node of the tree. Range: [0, inf). Default is 0
- Returns
- minSplitLoss value
◆ getObservationsPerTreeFraction()
double getObservationsPerTreeFraction |
( |
| ) |
|
Returns fraction of observations used for a training of one tree, 0 to 1. Default is 1 (sampling without replacement)
- Returns
- Fraction of observations
◆ getShrinkage()
Returns shrinkage parameter (learning rate) of the training procedure Scales the contribution of each tree by a factor (0, 1]. Default is 0.3
- Returns
- Shrinkage value
◆ getSplitMethod()
Returns split finding method of the gradient boosted trees training algorithm
- Returns
- Split finding method
◆ setEngine()
Sets the engine to be used by the algorithm
- Parameters
-
engine | to be used by the algorithm |
◆ setFeaturesPerNode()
void setFeaturesPerNode |
( |
long |
value | ) |
|
Sets the number of features tried as possible splits by gradient boosted trees classification training algorithm. If 0 then sqrt(p) is used, where p is the total number of features.
- Parameters
-
◆ setLambda()
void setLambda |
( |
double |
value | ) |
|
Sets lambda parameter value
- Parameters
-
◆ setMaxBins()
void setMaxBins |
( |
long |
value | ) |
|
Sets maximal number of discrete bins to bucket continuous features. Used with 'inexact' split finding method only. Default is 256. Increasing the number results in higher computation costs
- Parameters
-
value | Maximal number of discrete bins to bucket continuous features |
◆ setMaxIterations()
void setMaxIterations |
( |
long |
n | ) |
|
Sets maximal number of iterations for the gradient boosted trees training algorithm
- Parameters
-
◆ setMaxTreeDepth()
void setMaxTreeDepth |
( |
long |
value | ) |
|
Sets maximal tree depth. Default is 0 (unlimited)
- Parameters
-
◆ setMinBinSize()
void setMinBinSize |
( |
long |
value | ) |
|
Sets minimal number of observations in a bin. Default is 5 Used with 'inexact' split finding method only.
- Parameters
-
value | Minimal number of observations in a bin |
◆ setMinObservationsInLeafNode()
void setMinObservationsInLeafNode |
( |
long |
value | ) |
|
Sets minimal number of samples per node. Default is 1
- Parameters
-
value | Minimal number of samples |
◆ setMinSplitLoss()
void setMinSplitLoss |
( |
double |
value | ) |
|
Sets minSplitLoss parameter value
- Parameters
-
◆ setObservationsPerTreeFraction()
void setObservationsPerTreeFraction |
( |
double |
value | ) |
|
Sets fraction of observations used for a training of one tree, 0 to 1
- Parameters
-
value | Fraction of observations |
◆ setShrinkage()
void setShrinkage |
( |
double |
value | ) |
|
Sets shrinkage parameter value
- Parameters
-
◆ setSplitMethod()
Sets split finding method of the gradient boosted trees training algorithm
- Parameters
-
splitMethod | Split finding method |
The documentation for this class was generated from the following file:
- gbt/classification/training/Parameter.java