Parameter (Java Statistical Analysis Tool 0.0.8 API)

java.lang.Object
- jsat.parameters.Parameter

All Implemented Interfaces:

Serializable

Direct Known Subclasses:

BooleanParameter, DoubleParameter, IntParameter, MetricParameter, ObjectParameter
```
public abstract class Parameter
extends Object
implements Serializable
```
This interface provides a programmable manner in which the parameters of an algorithm may be altered and adjusted.

Author:

Edward Raff

See Also:

Serialized Form

Nested Class Summary

Nested Classes
Modifier and Type	Class and Description
`static interface`	`Parameter.ParameterHolder` Adding this annotation to a field tells the `getParamsFromMethods(java.lang.Object)` method to search this object recursively for more parameter get/set pairs. Placing this annotation on a `Collection` will cause the search to be done recursively over each item in the collection.
`static interface`	`Parameter.WarmParameter` Adding this annotation to a method tells the `getParamsFromMethods(java.lang.Object)` method to consider the Parameter object generated for that method a preferred parameter for warm starting.

Constructor Summary

Constructors
Constructor and Description

Parameter()

Constructors
Constructor and Description
`Parameter()`

Method Summary

All Methods Static Methods Instance Methods Abstract Methods Concrete Methods
Modifier and Type	Method and Description
`boolean`	`equals(Object obj)`
`abstract String`	`getASCIIName()` Returns the name of this parameter using only valid ACII characters.
`String`	`getName()` Returns the display name of this parameter.
`static List<Parameter>`	`getParamsFromMethods(Object obj)` Given an object, this method will use reflection to automatically find getter and setter method pairs, and create Parameter object for each getter setter pair. Getters are found by searching for no argument methods that start with "get" or "is".
`abstract String`	`getValueString()` Returns a string indicating the value currently held by the Parameter
`int`	`hashCode()`
`boolean`	`isWarmParameter()` If `true`, that means this parameter is the preferred parameter to be altered when warm starting from a previous version of the same class.
`boolean`	`preferredLowToHigh()` This value is meaningless if `isWarmParameter()` is `false` , and by default returns `false`.
`boolean`	`requiresRetrain()` Some variables of a learning method may be adjustable without having to re-train the whole data set.
`static Map<String,Parameter>`	`toParameterMap(List<Parameter> params)` Creates a map of all possible parameter names to their corresponding object.
`String`	`toString()`

Methods inherited from class java.lang.Object
clone, finalize, getClass, notify, notifyAll, wait, wait, wait

- Constructor Detail
  - Parameter
```
public Parameter()
```
- Method Detail
  - requiresRetrain
```
public boolean requiresRetrain()
```
    Some variables of a learning method may be adjustable without having to re-train the whole data set. false is returned if this is such a parameter, true if the learning method will need to be retrained after the parameter has changed.
    
    By default, this method returns true unless overwritten, as it is always safe to retrain the classifier if a parameter was changed.
    
    Returns:
    
    true if changing this parameter requires a re-training of the algorithm, or false if no-retraining is needed to take effect.
  - isWarmParameter
```
public boolean isWarmParameter()
```
    If true, that means this parameter is the preferred parameter to be altered when warm starting from a previous version of the same class. Being the preferred parameter means that using warms started training on a sequence of changes to this parameter should be faster than simply training normally for every combination of values.
    
    If more than one parameter would have this impact on training, the one that has the largest and most consistent impact should be selected.
    
    By default, this method will return false.
    
    Returns:
    
    true if warm starting on changes in this parameter has a significant impact on training speed, false otherwise.
  - preferredLowToHigh
```
public boolean preferredLowToHigh()
```
    This value is meaningless if isWarmParameter() is false , and by default returns false.
    
    This should return the preferred order of warm start value training in the progression of the warm parameter, either from low values to high values (ie: the model being trained has a higher value for this parameter than the warm model being trained from).
    
    Returns:
    
    true if warm starting on this parameter should occur from low values to high values, and false if it should go from high values to low.
  - getASCIIName
```
public abstract String getASCIIName()
```
    Returns the name of this parameter using only valid ACII characters.
    
    Returns:
    
    the ACII name
  - getName
```
public String getName()
```
    Returns the display name of this parameter. By default, this returns the ASCII name of the parameter. If one exists, a name using Unicode characters may be returned instead.
    
    Returns:
    
    the name of this parameter
  - toString
```
public String toString()
```
    Overrides:
    
    toString in class Object
  - hashCode
```
public int hashCode()
```
    Overrides:
    
    hashCode in class Object
  - getValueString
```
public abstract String getValueString()
```
    Returns a string indicating the value currently held by the Parameter
    
    Returns:
    
    a string representation of the parameter's value
  - equals
```
public boolean equals(Object obj)
```
    Overrides:
    
    equals in class Object
  - toParameterMap
```
public static Map<String,Parameter> toParameterMap(List<Parameter> params)
```
    Creates a map of all possible parameter names to their corresponding object. No two parameters may have the same name.
    
    Parameters:
    
    params - the list of parameters to create a map for
    
    Returns:
    
    a map of string names to their parameters
    
    Throws:
    
    RuntimeException - if two parameters have the same name
  - getParamsFromMethods
```
public static List<Parameter> getParamsFromMethods(Object obj)
```
    Given an object, this method will use reflection to automatically find getter and setter method pairs, and create Parameter object for each getter setter pair.
    Getters are found by searching for no argument methods that start with "get" or "is". Setters are found by searching for one argument methods that start with "set". A getter and setter are a pair only if everything after the prefix is the same in the method's name, and the return type of the getter is the same class as the argument for the setter.
    Current types supported are:
    - integer
    - doubles
    - booleans
    - Kernel Functions
    - Distance Metrics
    - Enums
    Parameters:
    
    obj -
    
    Returns:
    
    a list of parameter objects generated from the given object

Class Parameter

Nested Class Summary

Constructor Summary

Method Summary

Methods inherited from class java.lang.Object

Constructor Detail

Parameter

Method Detail

requiresRetrain

isWarmParameter

preferredLowToHigh

getASCIIName

getName

toString

hashCode

getValueString

equals

toParameterMap

getParamsFromMethods