Package com.bayesserver.analysis

Class RegressionStatistics

java.lang.Object

com.bayesserver.analysis.RegressionStatistics

public final class RegressionStatistics
extends Object

Calculates statistics for a network which is used to predict continuous values (regression).

Method Summary

Modifier and Type	Method	Description
static RegressionStatistics	create(DataReaderCommand readerCommand, String actual, String predicted)	Initializes a new instance of the RegressionStatistics class.
static RegressionStatistics	create(DataReaderCommand readerCommand, String actual, String predicted, String caseWeight)	Initializes a new instance of the RegressionStatistics class.
Double	getMeanAbsoluteError()	Gets the mean absolute error (MAE), which is a common measure used to determine how close predictions are to the actual values.
Double	getMeanActual()	Gets the mean of the actual column.
Double	getMeanSquaredError()	Gets the mean squared error (MSE), which is a common measure used to determine how close predictions are to the actual values.
Double	getRMSE()	Gets the root mean squared error (RMSE).
Double	getRSquared()	Gets the R squared value (Coefficient of determination).
Double	getSumAbsoluteError()	Gets the sum absolute error (SAE).
Double	getSumSquaredError()	Gets the sum of squared errors (SSE), which is a common measure used to determine how close predictions are to the actual values.
double	getSupport()	Gets the support/weight for the values used to calculate the statistics.
Double	getVarianceActual()	Gets the variance of the actual column.

Methods inherited from class java.lang.Object

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

Method Detail

create

public static RegressionStatistics create(DataReaderCommand readerCommand,
                                          String actual,
                                          String predicted)

Initializes a new instance of the RegressionStatistics class.

Parameters:

readerCommand - Returns the data containing the actual values and predicted values.

actual - The name of the data column containing the actual value.

predicted - The name of the data column containing the predicted value.

create

public static RegressionStatistics create(DataReaderCommand readerCommand,
                                          String actual,
                                          String predicted,
                                          String caseWeight)

Initializes a new instance of the RegressionStatistics class.

Parameters:

readerCommand - Returns the data containing the actual values and predicted values.

actual - The name of the data column containing the actual value.

predicted - The name of the data column containing the predicted value.

caseWeight - The case weight, if any. Can be null.

getMeanAbsoluteError

public Double getMeanAbsoluteError()

Gets the mean absolute error (MAE), which is a common measure used to determine how close predictions are to the actual values. The mean absolute error is the average of the absolute errors.

getSumAbsoluteError

public Double getSumAbsoluteError()

Gets the sum absolute error (SAE). The sum of the absolute errors.

getMeanSquaredError

public Double getMeanSquaredError()

Gets the mean squared error (MSE), which is a common measure used to determine how close predictions are to the actual values. The mean squared error is the average of the squared errors.

getSumSquaredError

public Double getSumSquaredError()

Gets the sum of squared errors (SSE), which is a common measure used to determine how close predictions are to the actual values.

getRMSE

public Double getRMSE()

Gets the root mean squared error (RMSE).

getMeanActual

public Double getMeanActual()

Gets the mean of the actual column.

getVarianceActual

public Double getVarianceActual()

Gets the variance of the actual column.

getRSquared

public Double getRSquared()

Gets the R squared value (Coefficient of determination).

getSupport

public double getSupport()

Gets the support/weight for the values used to calculate the statistics. (May not be a whole number, if original weights are not).