com.eignex.kumulant/stat/regression/glm/UnivariateRegressionResult

UnivariateRegressionResult

@Serializable

@SerialName(value = "UnivariateRegressionResult")

data class UnivariateRegressionResult(val penalty: Penalty, val totalWeights: Double, val slope: Double, val intercept: Double, val sse: Double, val sxy: Double, val x: VarianceResult, val y: VarianceResult) : Result, HasSlope, HasRegression(source)

Fitted univariate least-squares regression y = slope * x + intercept. Carries the marginal x / y variances and the raw weighted cross-deviation sxy so the result round-trips losslessly under merge regardless of penalty.

Constructors

UnivariateRegressionResult

constructor(penalty: Penalty, totalWeights: Double, slope: Double, intercept: Double, sse: Double, sxy: Double, x: VarianceResult, y: VarianceResult)(source)

Properties

correlation

val correlation: Double(source)

Pearson correlation of the x and y streams, sxy / sqrt(sxx * syy).

Computed from the raw cross-deviation rather than from 1 - sse/sst, because sse is evaluated at the penalised slope: under Penalty.L1 with a lambda above the soft-thresholding cut the slope is exactly zero, making sse == sst and yielding a correlation of zero for perfectly correlated data. This is a property of the two streams and so is independent of penalty.

covariance

val covariance: Double(source)

Weighted covariance sxy / totalWeights.

intercept

open override val intercept: Double(source)

Fitted intercept c.

penalty

val penalty: Penalty(source)

slope

open override val slope: Double(source)

Fitted slope coefficient m.

sse

open override val sse: Double(source)

Sum of squared errors (residuals). Under non-identity GLMs this is the deviance.

sst

open override val sst: Double(source)

Sum of squared deviations from the running mean: Sum (x - mean)^2 * weight.

Implementors must override this or variance; the two are defined in terms of each other, so overriding neither recurses until the stack runs out. Every Result in the library overrides sst, which is the cheaper of the two to carry.

bias

open override val bias: Double

Fitted bias / intercept term.

featureSize

open override val featureSize: Int

Number of features in weights.

isEmpty

open val isEmpty: Boolean

True when no observation has been folded in, so every other field is a placeholder.

mse

open val mse: Double

Mean squared error: sse / totalWeights. Zero on an empty stream.

rmse

open val rmse: Double

Root mean squared error: sqrt([mse]).

rSquared

open val rSquared: Double

Coefficient of determination 1 - sse/sst. Zero when sst is zero.

sampleStdDev

open val sampleStdDev: Double

Unbiased sample standard deviation: sqrt([sampleVariance]).

sampleVariance

open val sampleVariance: Double

Unbiased sample variance: [sst] / ([totalWeights] - 1). Zero when totalWeights <= 1.

ssr

open val ssr: Double

Sum of squares due to regression: sst - sse.

stdDev

open val stdDev: Double

Population standard deviation: sqrt([variance]).

variance

open val variance: Double

Population variance: [sst] / [totalWeights]. Zero on an empty stream.

weights

open override val weights: F64VectorView

Fitted weight per feature, indexed by the same i as the input x[i].

Functions

predict

open fun predict(x: F64VectorView): Double

Evaluate the linear predictor at x: bias + Sum_i weights[i] * x[i]. For Gaussian regression this is the prediction; for non-identity GLMs this is the linear predictor pre-link.

open fun predict(x: Double): Double

Evaluate the fitted line at x: slope * x + intercept.