SharpBCI.Predictor Class Reference

An IPredictor which uses a 3-dimensional loci of points in the form of an array of EEGEvent's to classify EEG data Uses nearest neighbor predictions with distance computed by the abstract Compute(double[] x, double[] y) function Predictions occur by computing the distance between the incoming sample and all training samples, pulls the k nearest neighbors, and determines the final prediction. More...

Inheritance diagram for SharpBCI.Predictor:

Public Member Functions
	Predictor (int channels, int k, double thresholdProb, EEGDataType[] bands)
void	AddTrainingData (int id, EEGEvent[] events)
void	ClearTrainingData ()
	Clears all training data stored within the predictor Essentially a reset of the entire prediction system Use between changing environments and/or participants. More...
void	SetChannelWeights (double[] newWeights)
void	SetBandWeights (double[] newWeights)
int	Predict (EEGEvent[] events)
List< KeyValuePair< int, double > >	ComputeDistances (double[][] data)
int	Vote (List< KeyValuePair< int, double >> distances)
Public Member Functions inherited from SharpBCI.IPredictor< T >
void	AddTrainingData (int label, T data)
	Add data to labeled training data set. More...
int	Predict (T test)
	Makes a prediction from the underlying model based off the new data. More...

Public Attributes
const int	NO_PREDICTION = -1
	see AggregatePredictionPipeable.NO_PREDICTION More...

Protected Member Functions
abstract double	Compute (double[] x, double[] y)

Protected Attributes
readonly Dictionary < EEGDataType, int >	bandLookup = new Dictionary<EEGDataType, int>()
	Dictionary which maps EEGDataTypes to integers for easy indexing into the buffer. More...
readonly Dictionary< int, List < double[][]> >	trainingData = new Dictionary<int, List<double[][]>>()
	Dictionary storing all training data. More...
readonly EEGDataType[]	bands
	List of EEGDataTypes that the predictor expects for its training and predictions. More...
readonly int	channels
	Number of channels on the headset, as reported by. More...
readonly int	k
	A non-negative non-zero integer representing the number of results from KNN algorithm. More...
readonly double	thresholdProb
	A double between 0-1 indicating the probability threshhold below which predictions will be thrown out. More...
double[]	bandWeights
double[]	channelWeights

Detailed Description

An IPredictor which uses a 3-dimensional loci of points in the form of an array of EEGEvent's to classify EEG data Uses nearest neighbor predictions with distance computed by the abstract Compute(double[] x, double[] y) function Predictions occur by computing the distance between the incoming sample and all training samples, pulls the k nearest neighbors, and determines the final prediction.

Constructor & Destructor Documentation

SharpBCI.Predictor.Predictor	(	int	channels,
		int	k,
		double	thresholdProb,
		EEGDataType[]	bands
	)

Member Function Documentation

void SharpBCI.Predictor.AddTrainingData	(	int	id,
		EEGEvent[]	events
	)

void SharpBCI.Predictor.ClearTrainingData

(

)

Clears all training data stored within the predictor Essentially a reset of the entire prediction system Use between changing environments and/or participants.

Implements SharpBCI.IPredictor< T >.

abstract double SharpBCI.Predictor.Compute ( double[]  x, double[]  y  )

protectedpure virtual

Implemented in SharpBCI.AggregateKNNDTWPredictor, and SharpBCI.AggregateKNNCorrelationPredictor.

List<KeyValuePair<int, double> > SharpBCI.Predictor.ComputeDistances

(

double

data[][]

)

int SharpBCI.Predictor.Predict

(

EEGEvent[]

events

)

void SharpBCI.Predictor.SetBandWeights

(

double[]

newWeights

)

void SharpBCI.Predictor.SetChannelWeights

(

double[]

newWeights

)

int SharpBCI.Predictor.Vote

(

List< KeyValuePair< int, double >>

distances

)

Member Data Documentation

readonly Dictionary<EEGDataType, int> SharpBCI.Predictor.bandLookup = new Dictionary<EEGDataType, int>()

protected

Dictionary which maps EEGDataTypes to integers for easy indexing into the buffer.

readonly EEGDataType [] SharpBCI.Predictor.bands

protected

List of EEGDataTypes that the predictor expects for its training and predictions.

double [] SharpBCI.Predictor.bandWeights

protected

readonly int SharpBCI.Predictor.channels

protected

Number of channels on the headset, as reported by.

See Also

SharpBCIAdapter

double [] SharpBCI.Predictor.channelWeights

protected

readonly int SharpBCI.Predictor.k

protected

A non-negative non-zero integer representing the number of results from KNN algorithm.

https://en.wikipedia.org/wiki/K-nearest_neighbors_algorithm

const int SharpBCI.Predictor.NO_PREDICTION = -1

see AggregatePredictionPipeable.NO_PREDICTION

readonly double SharpBCI.Predictor.thresholdProb

protected

A double between 0-1 indicating the probability threshhold below which predictions will be thrown out.

readonly Dictionary<int, List<double[][]> > SharpBCI.Predictor.trainingData = new Dictionary<int, List<double[][]>>()

protected

Dictionary storing all training data.

Key is the label of the value's data Value is a list of data associated with this label

---

The documentation for this class was generated from the following file:

• /home/travis/build/berkona/SharpBCI/SharpBCI/AggregatePredictors.cs