jsat.driftdetectors

Class DDM<V>

java.lang.Object

jsat.driftdetectors.BaseDriftDetector<V>

jsat.driftdetectors.DDM<V>

All Implemented Interfaces:

Serializable, Cloneable

public class DDM<V>
extends BaseDriftDetector<V>

DDM (Drift Detection Method) is a drift detector for binary events, and is meant to detect decreases in the success rate over time. As such it will not inform of any positive drift.

This drift detector supports a warning state, and will not begin to store the object history until a warning state begins. If the warning state ends before a detection of drift occurs, the history will be dropped.
See: Gama, J., Medas, P., Castillo, G.,&Rodrigues, P. (2004). Learning with Drift Detection. In A. C. Bazzan&S. Labidi (Eds.), Advances in Artificial Intelligence – SBIA 2004 (pp. 286–295). Springer Berlin Heidelberg. doi:10.1007/978-3-540-28645-5_29

Author:

Edward Raff

See Also:

Serialized Form

Field Summary

Fields inherited from class jsat.driftdetectors.BaseDriftDetector

drifting, driftStart, history, maxHistory, time, warning

Constructor Summary

Constructors

Constructor and Description
DDM() Creates a new DDM drift detector using the default warning and drift thresholds of 2 and 3 respectively.
DDM(DDM<V> toCopy) Copy constructor
DDM(double warningThreshold, double driftThreshold) Creates a new DDM drift detector

Method Summary

Modifier and Type	Method and Description
boolean	addSample(boolean trial, V obj) Adds a new boolean trial to the detector, with the goal of detecting when the number of successful trials (true) drifts to a new value.
boolean	addSample(double value, V obj) Adds a new point to the drift detector.
DDM<V>	clone()
void	driftHandled() This method should be called once the drift is handled.
double	getDriftThreshold() Returns the threshold multiple for controlling the false positive / negative rate on detecting changes.
double	getSuccessRate() Returns the current estimate of the success rate (number of true inputs) for the model.
double	getWarningThreshold() Returns the threshold multiple for controlling the false positive / negative rate on detecting changes.
void	setDriftThreshold(double driftThreshold) Sets the multiplier on the standard deviation that must be exceeded to recognize the change as a drift.
void	setWarningThreshold(double warningThreshold) Sets the multiplier on the standard deviation that must be exceeded to initiate a warning state.

Methods inherited from class jsat.driftdetectors.BaseDriftDetector

addToHistory, clearHistory, getDriftAge, getDriftedHistory, getMaxHistory, isDrifting, isWarning, setMaxHistory

Methods inherited from class java.lang.Object

equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

DDM

public DDM()

Creates a new DDM drift detector using the default warning and drift thresholds of 2 and 3 respectively.

DDM

public DDM(double warningThreshold,
           double driftThreshold)

Creates a new DDM drift detector

Parameters:

warningThreshold - the threshold for starting a warning state

driftThreshold - the threshold for recognizing a drift

DDM

public DDM(DDM<V> toCopy)

Copy constructor

Parameters:

toCopy - the object to copy

Method Detail

getSuccessRate

public double getSuccessRate()

Returns the current estimate of the success rate (number of true inputs) for the model.

Returns:

the current estimate of the success rate

addSample

public boolean addSample(boolean trial,
                         V obj)

Adds a new boolean trial to the detector, with the goal of detecting when the number of successful trials (true) drifts to a new value. This detector begins storing a history of the obj inputs only once it has entered a warning state.
This detector is specifically meant to detect drops in the success rate, and will not cause any warning or drift detections for increases in the success rate.

Parameters:

trial - the result of the trial

obj - the object to associate with the trial

Returns:

true if we are in a warning or drift state, false if we are not

setWarningThreshold

public void setWarningThreshold(double warningThreshold)

Sets the multiplier on the standard deviation that must be exceeded to initiate a warning state. Once in the warning state, DDM will begin to collect a history of the inputs
Increasing the warning threshold makes it take longer to start detecting a change, but reduces false positives.
If the warning threshold is set above the setDriftThreshold(double), the drift state will not occur until the warning state is reached, and the warning state will be skipped.

Parameters:

warningThreshold - the positive multiplier threshold for starting a warning state

getWarningThreshold

public double getWarningThreshold()

Returns the threshold multiple for controlling the false positive / negative rate on detecting changes.

Returns:

the threshold multiple for controlling warning detection

setDriftThreshold

public void setDriftThreshold(double driftThreshold)

Sets the multiplier on the standard deviation that must be exceeded to recognize the change as a drift.
Increasing the drift threshold makes it take longer to start detecting a change, but reduces false positives.

Parameters:

driftThreshold - the positive multiplier threshold for detecting a drift

getDriftThreshold

public double getDriftThreshold()

Returns the threshold multiple for controlling the false positive / negative rate on detecting changes.

Returns:

the threshold for controlling drift detection

addSample

public boolean addSample(double value,
                         V obj)

Description copied from class: BaseDriftDetector

Adds a new point to the drift detector. If an escalation in state occurs, true will be returned. A change of state could be either drift occurring BaseDriftDetector.isDrifting() or a warning state BaseDriftDetector.isWarning(). If the detector was in a warning state and then goes back to normal, false will be returned.

For binary (true / false) drift detectors, value will be considered false if and only if its value is equal to zero. Any non zero value will be treated as true

Objects added with the value may or may not be added to the history, the behavior is algorithm dependent. Some may always add it to the history, while others will only begin collecting history once a warning state occurs.

Specified by:

addSample in class BaseDriftDetector<V>

Parameters:

value - the numeric value to add to the drift detector

obj - the object associated with this value. It may or may not be stored in the detectors history

Returns:

true if a drift has or may be starting.

driftHandled

public void driftHandled()

Description copied from class: BaseDriftDetector

This method should be called once the drift is handled. Once done, this method will clear the flags and prepare the detector to continue tracking drift again.
By using this method, one can continue to track multiple future drift changes without having to feed the history data (which may be incomplete) into a new detector object.

Overrides:

driftHandled in class BaseDriftDetector<V>

clone

public DDM<V> clone()

Specified by:

clone in class BaseDriftDetector<V>