BaseTreeSettings

All Superinterfaces:

Verifiable

All Known Subinterfaces:

RegressionTreeSettings, TreeSettings
```
public interface BaseTreeSettings
extends Verifiable
```
General algorithm settings for tree algorithm. It is responsible for providing algorithm settings for the proper algorithm.

Method Summary

Methods
Modifier and Type	Method and Description
`double`	`getConfidenceFactor()` Confidence factor for estimating error made by tree for pruning purposes.
`double`	`getHomogenousDensity()` The smallest density of any target in node to justify changing it into leaf.
`int`	`getMaxNodeRank()` Maximum number of children for each node.
`int`	`getMaxNumberOfBestPredicates()`
`int`	`getMaxTreeDepth()` Maximum depth of the tree. 0 means no height limit.
`double`	`getMinLeafSize()` What is the minimum leaf size.
`double`	`getMinLeafSizePrc()` What is the minimum leaf size relatively to the whole data set size.
`double`	`getMinNodeSize()` Minimum number of samples for a node.
`double`	`getMinNodeSizePrc()` Minimal node size calculated as percent of the training set size.
`double`	`getMinSplitEvaluationGain()` The minimum gain in split evaluation value, to continue splitting a node.
`int`	`getNumberOfHistogramQuantiles()` Number of histogram quantiles in split editor
`int`	`getSplitSampling()`
`boolean`	`isCalculateSplitStatistics()` Calculate split statistics
`boolean`	`isNodeRules()`
`boolean`	`isNullTestAllowed()` Experimental feature.
`boolean`	`isPruneTree()` Decides whether tree will be pruned after building.
`boolean`	`isPruneValidation()`
`boolean`	`isSmoothHistograms()` If true, histograms in split statistics will be smooth, using kernel density estimation method
`void`	`setCalculateSplitStatistics(boolean set)` Calculate split statistics
`void`	`setConfidenceFactor(double factor)` Confidence factor for estimating error made by tree for pruning purposes.
`void`	`setHomogenousDensity(double maxPValue)` The smallest density of any target in node to justify changing it into leaf.
`void`	`setMaxNodeRank(int maxSplits)` Maximum number of children nodes for a node.
`void`	`setMaxNumberOfBestPredicates(int maxNumberOfPredicates)` Maximum number of best predicates computed in the building phase of the algorithm.
`void`	`setMaxTreeDepth(int maxDepth)` The maximum tree depth.
`void`	`setMinLeafSize(double minLeafSize)` Sets the minimum size of a tree leaf - a node split is performed only if every leaf has at least that number of observations.
`void`	`setMinLeafSizePrc(double minLeafSizePrc)` Sets the minimum size of a tree leaf relatively to the whole data set size - a node split is performed only if every leaf has at least that number of observations.
`void`	`setMinNodeSize(double size)` The minimal number of samples for a node.
`void`	`setMinNodeSizePrc(double prc)` Minimal node size calculated as percent of the training set size.
`void`	`setMinSplitEvaluationGain(double minImpurity)` The minimum gain in split evaluation value, to continue splitting a node.
`void`	`setNodeRules(boolean b)`
`void`	`setNullTestAllowed(boolean allowed)` Experimental feature.
`void`	`setNumberOfHistogramQuantiles(int quantiles)` Number of histogram quantiles in split editor
`void`	`setPruneTree(boolean prune)` Decides whether tree will be pruned after building.
`void`	`setPruneValidation(boolean b)`
`void`	`setSmoothHistograms(boolean smooth)` If true, histograms in split statistics will be smooth, using kernel density estimation method
`void`	`setSplitSampling(int s)`

Methods inherited from interface biz.sc.mm.dm.base.Verifiable
verify

- Method Detail
  - setMaxTreeDepth
```
void setMaxTreeDepth(int maxDepth)
```
    The maximum tree depth. Constructed tree in building phase, before pruning, will contain number of levels specified by this parameter.Please note, that the constraint holds in the building phase, and might not hold after pruning. It implies, that after pruning the tree will contain at most the specified number of levels. The parameter will affect the tree size, model size, and overall time needed to perform tree building.
    
    Parameters:
    maxDepth - Maximum tree depth. Must be >= 0. If 0 is set, no there are no restriction to the tree depth. Default value is 20.
  - getMaxTreeDepth
```
int getMaxTreeDepth()
                    throws GSException
```
    Maximum depth of the tree. 0 means no height limit.
    
    Returns:
    maximum depth of the tree
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
    See Also:
    setMaxTreeDepth(int)
  - setMinNodeSize
```
void setMinNodeSize(double size)
                    throws GSException
```
    The minimal number of samples for a node. Splitting of given node is continued only, if all sub-nodes satisfies this condition. In result, constructed tree does not have any nodes with number of samples less than Min Node Size value. Please note, that the constraint holds in the building phase, and might not hold after pruning. The parameter greatly influences size of the result tree, removing all leaves with size less, than a given value.
    
    Parameters:
    size - minimal number of samples for a node. Real number >=0. If set to 0, there is no restriction to the node size. Default value 0.0.
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
  - getMinNodeSize
```
double getMinNodeSize()
                      throws GSException
```
    Minimum number of samples for a node.
    
    Returns:
    minimum number of samples for a node
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
    See Also:
    setMinNodeSize(double)
  - getMinSplitEvaluationGain
```
double getMinSplitEvaluationGain()
                                 throws GSException
```
    The minimum gain in split evaluation value, to continue splitting a node.
    
    Returns:
    the minimum gain in split evaluation value, to continue splitting a node
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
    See Also:
    setMinSplitEvaluationGain(double)
  - setMinSplitEvaluationGain
```
void setMinSplitEvaluationGain(double minImpurity)
                               throws GSException
```
    The minimum gain in split evaluation value, to continue splitting a node. The value is computed depending on function defined in splitEvaluation parameter. Only splits with evaluation value greater than or equal the minimal split evaluation gain, are considered by algorithm to be set as a node split. In other case, the leaf is created.
    
    Parameters:
    minImpurity - The minimum gain in split evaluation value, to continue splitting a node. Real number [0.0, 1.0]. Default value 0.0.
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
  - getMaxNodeRank
```
int getMaxNodeRank()
                   throws GSException
```
    Maximum number of children for each node. Influences nominal tests only.
    
    Returns:
    Maximum number of children for each node.
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
    See Also:
    setMaxNodeRank(int)
  - setMaxNodeRank
```
void setMaxNodeRank(int maxSplits)
                    throws GSException
```
    Maximum number of children nodes for a node. Parameter influences only splits for nominal attribute types. Numeric tests are always binary. This property has double meaning, depending whether attribute value grouping is on or off:
    - value grouping is ON
      Algorithm takes two types of tests for nominal attributes: binary, and grouped. It starts value grouping, every time joining two best values together, decreasing number of values in split. The process iterates, until every value joining decreases the assessment value. Result of grouping is split with some number of possible results. If the number is less than max node rank value, than grouped split is used. In other way binary split is used.
    - value grouping is OFF
      Algorithm takes two types of tests for nominal attributes: binary or multi-value. If number of values for the attribute is less than max node rank value, than multi-value is used. If not, than binary is used.
    Parameters:
    maxSplits - Maximum number of children for each node. Integer number >= 2. Default value is 2.
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
  - getHomogenousDensity
```
double getHomogenousDensity()
                            throws GSException
```
    The smallest density of any target in node to justify changing it into leaf. A node with a target value probability exceeding the maximum value is considered to be homogeneous, i.e., no further splitting is required. This node becomes terminal (a leaf node). "Value must be between 0 and 1.
    
    Returns:
    The smallest density of any target in node to justify changing it into leaf.
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
    See Also:
    setMaxNodeRank(int)
  - setHomogenousDensity
```
void setHomogenousDensity(double maxPValue)
                          throws GSException
```
    The smallest density of any target in node to justify changing it into leaf. A node with a target value probability exceeding the maximum value is considered to be homogeneous, i.e., no further splitting is required. This node becomes terminal (a leaf node). "Value must be between 0 and 1.
    
    Parameters:
    maxPValue - The smallest density of any target in node to justify changing it into leaf. Real number (0.0, 1.0]. Default value is 1.0.
    
    Throws:
    
    biz.sc.gornik.common.GSException
    
    GSException
  - setPruneTree
```
void setPruneTree(boolean prune)
```
    Decides whether tree will be pruned after building.
    
    Parameters:
    prune - Decides whether tree will be pruned after building. Default value is true.
  - isPruneTree
```
boolean isPruneTree()
```
    Decides whether tree will be pruned after building.
    
    Returns:
    Boolean value whether tree will be pruned after building.
    See Also:
    setPruneTree(boolean)
  - setConfidenceFactor
```
void setConfidenceFactor(double factor)
```
    Confidence factor for estimating error made by tree for pruning purposes. Used by Pessimistic Error Pruning, when pruneTree is set to 'yes'. Smaller the value, more often algorithm will replace subtree with leaf. Valid values are [0.0, 0.5). Default value is 0.25.
    
    Parameters:
    factor - Confidence factor value
  - getConfidenceFactor
```
double getConfidenceFactor()
```
    Confidence factor for estimating error made by tree for pruning purposes.
    
    Returns:
    Confidence factor value
    See Also:
    setConfidenceFactor(double)
  - getMaxNumberOfBestPredicates
```
int getMaxNumberOfBestPredicates()
```
  - setMaxNumberOfBestPredicates
```
void setMaxNumberOfBestPredicates(int maxNumberOfPredicates)
```
    Maximum number of best predicates computed in the building phase of the algorithm. Best predicates are shown in biz.sc.mm.dm.modelstats.tree.Node#getBestPredicates() property.
    
    Parameters:
    maxNumberOfPredicates - maximum number of best predicates. Valid values are integers greater or equal 1
  - isCalculateSplitStatistics
```
boolean isCalculateSplitStatistics()
```
    Calculate split statistics
    
    Returns:
  - setCalculateSplitStatistics
```
void setCalculateSplitStatistics(boolean set)
```
    Calculate split statistics
  - getNumberOfHistogramQuantiles
```
int getNumberOfHistogramQuantiles()
```
    Number of histogram quantiles in split editor
  - setNumberOfHistogramQuantiles
```
void setNumberOfHistogramQuantiles(int quantiles)
```
    Number of histogram quantiles in split editor
  - isSmoothHistograms
```
boolean isSmoothHistograms()
```
    If true, histograms in split statistics will be smooth, using kernel density estimation method
  - setSmoothHistograms
```
void setSmoothHistograms(boolean smooth)
```
    If true, histograms in split statistics will be smooth, using kernel density estimation method
  - getMinNodeSizePrc
```
double getMinNodeSizePrc()
```
    Minimal node size calculated as percent of the training set size. Inactive when set to 0
  - setMinNodeSizePrc
```
void setMinNodeSizePrc(double prc)
```
    Minimal node size calculated as percent of the training set size. Inactive when set to 0
  - setMinLeafSize
```
void setMinLeafSize(double minLeafSize)
```
    Sets the minimum size of a tree leaf - a node split is performed only if every leaf has at least that number of observations.
    
    Parameters:
    minLeafSize - The number of observations.
  - getMinLeafSize
```
double getMinLeafSize()
```
    What is the minimum leaf size.
    
    Returns:
    The number of observations.
    See Also:
    setMinLeafSize(double)
  - setMinLeafSizePrc
```
void setMinLeafSizePrc(double minLeafSizePrc)
```
    Sets the minimum size of a tree leaf relatively to the whole data set size - a node split is performed only if every leaf has at least that number of observations.
    
    Parameters:
    minLeafSizePrc - The number of observations.
  - getMinLeafSizePrc
```
double getMinLeafSizePrc()
```
    What is the minimum leaf size relatively to the whole data set size.
    
    Returns:
    Leaf size / data set size ratio in percents.
    See Also:
    setMinLeafSizePrc(double)
  - setNullTestAllowed
```
void setNullTestAllowed(boolean allowed)
```
    Experimental feature. If true, then split for attribute with null values is evaluated with lower value.
  - isNullTestAllowed
```
boolean isNullTestAllowed()
```
    Experimental feature. If true, then split for attribute with null values is evaluated with lower value.
  - setPruneValidation
```
void setPruneValidation(boolean b)
```
  - isPruneValidation
```
boolean isPruneValidation()
```
  - getSplitSampling
```
int getSplitSampling()
```
  - setSplitSampling
```
void setSplitSampling(int s)
```
  - isNodeRules
```
boolean isNodeRules()
```
  - setNodeRules
```
void setNodeRules(boolean b)
```

Interface BaseTreeSettings

Method Summary

Methods inherited from interface biz.sc.mm.dm.base.Verifiable

Method Detail

setMaxTreeDepth

getMaxTreeDepth

setMinNodeSize

getMinNodeSize

getMinSplitEvaluationGain

setMinSplitEvaluationGain

getMaxNodeRank

setMaxNodeRank

getHomogenousDensity

setHomogenousDensity

setPruneTree

isPruneTree

setConfidenceFactor

getConfidenceFactor

getMaxNumberOfBestPredicates

setMaxNumberOfBestPredicates

isCalculateSplitStatistics

setCalculateSplitStatistics

getNumberOfHistogramQuantiles

setNumberOfHistogramQuantiles

isSmoothHistograms

setSmoothHistograms

getMinNodeSizePrc

setMinNodeSizePrc

setMinLeafSize

getMinLeafSize

setMinLeafSizePrc

getMinLeafSizePrc

setNullTestAllowed

isNullTestAllowed

setPruneValidation

isPruneValidation

getSplitSampling

setSplitSampling

isNodeRules

setNodeRules