metrics

package
v0.0.0-...-8433fb5 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Sep 11, 2018 License: MIT Imports: 11 Imported by: 9

Documentation

Index

Constants

This section is empty.

Variables

This section is empty.

Functions

This section is empty.

Types

type AbsolutePearson 

type AbsolutePearson struct{}

AbsolutePearson measures the ROC AUC score.

func (AbsolutePearson) Apply 

func (ap AbsolutePearson) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply AbsolutePearson.

func (AbsolutePearson) BiggerIsBetter 

func (ap AbsolutePearson) BiggerIsBetter() bool

BiggerIsBetter method of AbsolutePearson.

func (AbsolutePearson) Classification 

func (ap AbsolutePearson) Classification() bool

Classification method of AbsolutePearson.

func (AbsolutePearson) NeedsProbabilities 

func (ap AbsolutePearson) NeedsProbabilities() bool

NeedsProbabilities method of AbsolutePearson.

func (AbsolutePearson) String 

func (ap AbsolutePearson) String() string

String method of AbsolutePearson.

type Accuracy 

type Accuracy struct{}

Accuracy measures the fraction of matches between true classes and predicted classes.

func (Accuracy) Apply 

func (acc Accuracy) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply Accuracy.

func (Accuracy) BiggerIsBetter 

func (acc Accuracy) BiggerIsBetter() bool

BiggerIsBetter method of Accuracy.

func (Accuracy) Classification 

func (acc Accuracy) Classification() bool

Classification method of Accuracy.

func (Accuracy) NeedsProbabilities 

func (acc Accuracy) NeedsProbabilities() bool

NeedsProbabilities method of Accuracy.

func (Accuracy) String 

func (acc Accuracy) String() string

String method of Accuracy.

type ConfusionMatrix 

type ConfusionMatrix map[float64]map[float64]float64

A ConfusionMatrix stores true positives (TP), true negatives (TN), false positives (FP) and false negatives (FN).

func MakeConfusionMatrix 

func MakeConfusionMatrix(yTrue, yPred, weights []float64) (ConfusionMatrix, error)

MakeConfusionMatrix returns a ConfusionMatrix from a slice of true classes and another slice of predicted classes.

func (ConfusionMatrix) Classes 

func (cm ConfusionMatrix) Classes() []float64

Classes returns a slice of classes included in a ConfusionMatrix. The result is ordered in ascending order.

func (ConfusionMatrix) FalseNegatives 

func (cm ConfusionMatrix) FalseNegatives(class float64) float64

FalseNegatives returns the number of times a class was wrongly not predicted.

func (ConfusionMatrix) FalsePositives 

func (cm ConfusionMatrix) FalsePositives(class float64) float64

FalsePositives returns the number of times a class was wrongly predicted.

func (ConfusionMatrix) NClasses 

func (cm ConfusionMatrix) NClasses() int

NClasses returns the number of classes in a ConfusionMatrix.

func (ConfusionMatrix) String 

func (cm ConfusionMatrix) String() string

String returns a string that can easily be read by a human in a terminal.

func (ConfusionMatrix) TrueNegatives 

func (cm ConfusionMatrix) TrueNegatives(class float64) float64

TrueNegatives returns the number of times a class was correctly not predicted.

func (ConfusionMatrix) TruePositives 

func (cm ConfusionMatrix) TruePositives(class float64) float64

TruePositives returns the number of times a class was correctly predicted.

type DiffMetric 

type DiffMetric interface {
	Metric
	Gradients(yTrue, yPred []float64) ([]float64, error)
}

A DiffMetric is a Metric that can compute element-wise gradients.

type F1 

type F1 struct {
	Class float64
}

F1 measures the F1-score.

func (F1) Apply 

func (f1 F1) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply F1.

func (F1) BiggerIsBetter 

func (f1 F1) BiggerIsBetter() bool

BiggerIsBetter method of F1.

func (F1) Classification 

func (f1 F1) Classification() bool

Classification method of F1.

func (F1) NeedsProbabilities 

func (f1 F1) NeedsProbabilities() bool

NeedsProbabilities method of F1.

func (F1) String 

func (f1 F1) String() string

String method of F1.

type LogLoss 

type LogLoss struct{}

LogLoss implementes logistic loss.

func (LogLoss) Apply 

func (ll LogLoss) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply LogLoss.

func (LogLoss) BiggerIsBetter 

func (ll LogLoss) BiggerIsBetter() bool

BiggerIsBetter method of LogLoss.

func (LogLoss) Classification 

func (ll LogLoss) Classification() bool

Classification method of LogLoss.

func (LogLoss) Gradients 

func (ll LogLoss) Gradients(yTrue, yPred []float64) ([]float64, error)

Gradients computes yPred[i] - yTrue[i].

func (LogLoss) NeedsProbabilities 

func (ll LogLoss) NeedsProbabilities() bool

NeedsProbabilities method of LogLoss.

func (LogLoss) String 

func (ll LogLoss) String() string

String method of LogLoss.

type MAE 

type MAE struct{}

MAE measures the mean absolute error (MAE).

func (MAE) Apply 

func (mae MAE) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply MAE.

func (MAE) BiggerIsBetter 

func (mae MAE) BiggerIsBetter() bool

BiggerIsBetter method of MAE.

func (MAE) Classification 

func (mae MAE) Classification() bool

Classification method of MAE.

func (MAE) Gradients 

func (mae MAE) Gradients(yTrue, yPred []float64) ([]float64, error)

Gradients computes sign(yPred[i] - yTrue[i]).

func (MAE) NeedsProbabilities 

func (mae MAE) NeedsProbabilities() bool

NeedsProbabilities method of MAE.

func (MAE) String 

func (mae MAE) String() string

String method of MAE.

type MSE 

type MSE struct{}

MSE measures the mean squared error (MSE).

func (MSE) Apply 

func (mse MSE) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply MSE.

func (MSE) BiggerIsBetter 

func (mse MSE) BiggerIsBetter() bool

BiggerIsBetter method of MSE.

func (MSE) Classification 

func (mse MSE) Classification() bool

Classification method of MSE.

func (MSE) Gradients 

func (mse MSE) Gradients(yTrue, yPred []float64) ([]float64, error)

Gradients computes 2 * (yPred[i] - yTrue[i]).

func (MSE) NeedsProbabilities 

func (mse MSE) NeedsProbabilities() bool

NeedsProbabilities method of MSE.

func (MSE) String 

func (mse MSE) String() string

String method of MSE.

type MacroF1 

type MacroF1 struct{}

MacroF1 measures the global F1 score.

func (MacroF1) Apply 

func (f1 MacroF1) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply MacroF1.

func (MacroF1) BiggerIsBetter 

func (f1 MacroF1) BiggerIsBetter() bool

BiggerIsBetter method of MacroF1.

func (MacroF1) Classification 

func (f1 MacroF1) Classification() bool

Classification method of MacroF1.

func (MacroF1) NeedsProbabilities 

func (f1 MacroF1) NeedsProbabilities() bool

NeedsProbabilities method of MacroF1.

func (MacroF1) String 

func (f1 MacroF1) String() string

String method of MacroF1.

type MacroPrecision 

type MacroPrecision struct{}

MacroPrecision measures the unweighted average precision across all classes. This does not take class imbalance into account.

func (MacroPrecision) Apply 

func (precision MacroPrecision) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply MacroPrecision.

func (MacroPrecision) BiggerIsBetter 

func (precision MacroPrecision) BiggerIsBetter() bool

BiggerIsBetter method of MacroPrecision.

func (MacroPrecision) Classification 

func (precision MacroPrecision) Classification() bool

Classification method of MacroPrecision.

func (MacroPrecision) NeedsProbabilities 

func (precision MacroPrecision) NeedsProbabilities() bool

NeedsProbabilities method of MacroPrecision.

func (MacroPrecision) String 

func (precision MacroPrecision) String() string

String method of MacroPrecision.

type MacroRecall 

type MacroRecall struct{}

MacroRecall measures the unweighted average recall across all classes. This does not take class imbalance into account.

func (MacroRecall) Apply 

func (recall MacroRecall) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply MacroRecall.

func (MacroRecall) BiggerIsBetter 

func (recall MacroRecall) BiggerIsBetter() bool

BiggerIsBetter method of MacroRecall.

func (MacroRecall) Classification 

func (recall MacroRecall) Classification() bool

Classification method of MacroRecall.

func (MacroRecall) NeedsProbabilities 

func (recall MacroRecall) NeedsProbabilities() bool

NeedsProbabilities method of MacroRecall.

func (MacroRecall) String 

func (recall MacroRecall) String() string

String method of MacroRecall.

type Metric 

type Metric interface {
	Apply(yTrue, yPred, weights []float64) (float64, error)
	Classification() bool
	BiggerIsBetter() bool
	NeedsProbabilities() bool
	String() string
}

A Metric metricuates the performance of a predictive model. yTrue, yPred, and weights should all have the same length. If weights is nil then uniform weights are used.

func ParseMetric 

func ParseMetric(name string, class int) (Metric, error)

ParseMetric returns a Metric from it's String representation.

type MicroF1 

type MicroF1 struct{}

MicroF1 measures the global F1 score.

func (MicroF1) Apply 

func (f1 MicroF1) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply MicroF1.

func (MicroF1) BiggerIsBetter 

func (f1 MicroF1) BiggerIsBetter() bool

BiggerIsBetter method of MicroF1.

func (MicroF1) Classification 

func (f1 MicroF1) Classification() bool

Classification method of MicroF1.

func (MicroF1) NeedsProbabilities 

func (f1 MicroF1) NeedsProbabilities() bool

NeedsProbabilities method of MicroF1.

func (MicroF1) String 

func (f1 MicroF1) String() string

String method of MicroF1.

type MicroPrecision 

type MicroPrecision struct{}

MicroPrecision measures the global precision by using the total true positives and false positives.

func (MicroPrecision) Apply 

func (precision MicroPrecision) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply MicroPrecision.

func (MicroPrecision) BiggerIsBetter 

func (precision MicroPrecision) BiggerIsBetter() bool

BiggerIsBetter method of MicroPrecision.

func (MicroPrecision) Classification 

func (precision MicroPrecision) Classification() bool

Classification method of MicroPrecision.

func (MicroPrecision) NeedsProbabilities 

func (precision MicroPrecision) NeedsProbabilities() bool

NeedsProbabilities method of MicroPrecision.

func (MicroPrecision) String 

func (precision MicroPrecision) String() string

String method of MicroPrecision.

type MicroRecall 

type MicroRecall struct{}

MicroRecall measures the global recall by using the total true positives and false negatives.

func (MicroRecall) Apply 

func (recall MicroRecall) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply MicroRecall.

func (MicroRecall) BiggerIsBetter 

func (recall MicroRecall) BiggerIsBetter() bool

BiggerIsBetter method of MicroRecall.

func (MicroRecall) Classification 

func (recall MicroRecall) Classification() bool

Classification method of MicroRecall.

func (MicroRecall) NeedsProbabilities 

func (recall MicroRecall) NeedsProbabilities() bool

NeedsProbabilities method of MicroRecall.

func (MicroRecall) String 

func (recall MicroRecall) String() string

String method of MicroRecall.

type Negative 

type Negative struct {
	Metric Metric
}

A Negative returns the negative output of a given Metric.

func (Negative) Apply 

func (neg Negative) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply Negative.

func (Negative) BiggerIsBetter 

func (neg Negative) BiggerIsBetter() bool

BiggerIsBetter method of Negative.

func (Negative) Classification 

func (neg Negative) Classification() bool

Classification method of Negative.

func (Negative) NeedsProbabilities 

func (neg Negative) NeedsProbabilities() bool

NeedsProbabilities method of Negative.

func (Negative) String 

func (neg Negative) String() string

String method of Negative.

type Precision 

type Precision struct {
	Class float64
}

Precision measures the fraction of times a class was correctly predicted.

func (Precision) Apply 

func (precision Precision) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply Precision.

func (Precision) BiggerIsBetter 

func (precision Precision) BiggerIsBetter() bool

BiggerIsBetter method of Precision.

func (Precision) Classification 

func (precision Precision) Classification() bool

Classification method of Precision.

func (Precision) NeedsProbabilities 

func (precision Precision) NeedsProbabilities() bool

NeedsProbabilities method of Precision.

func (Precision) String 

func (precision Precision) String() string

String method of Precision.

type R2 

type R2 struct{}

R2 measures the coefficient of determination.

func (R2) Apply 

func (r2 R2) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply R2.

func (R2) BiggerIsBetter 

func (r2 R2) BiggerIsBetter() bool

BiggerIsBetter method of R2.

func (R2) Classification 

func (r2 R2) Classification() bool

Classification method of R2.

func (R2) NeedsProbabilities 

func (r2 R2) NeedsProbabilities() bool

NeedsProbabilities method of R2.

func (R2) String 

func (r2 R2) String() string

String method of R2.

type RMSE 

type RMSE struct{}

RMSE measures the root mean squared error (RMSE).

func (RMSE) Apply 

func (rmse RMSE) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply RMSE.

func (RMSE) BiggerIsBetter 

func (rmse RMSE) BiggerIsBetter() bool

BiggerIsBetter method of RMSE.

func (RMSE) Classification 

func (rmse RMSE) Classification() bool

Classification method of RMSE.

func (RMSE) NeedsProbabilities 

func (rmse RMSE) NeedsProbabilities() bool

NeedsProbabilities method of RMSE.

func (RMSE) String 

func (rmse RMSE) String() string

String method of RMSE.

type ROCAUC 

type ROCAUC struct{}

ROCAUC measures the ROC AUC score.

func (ROCAUC) Apply 

func (rocauc ROCAUC) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply ROCAUC.

func (ROCAUC) BiggerIsBetter 

func (rocauc ROCAUC) BiggerIsBetter() bool

BiggerIsBetter method of ROCAUC.

func (ROCAUC) Classification 

func (rocauc ROCAUC) Classification() bool

Classification method of ROCAUC.

func (ROCAUC) NeedsProbabilities 

func (rocauc ROCAUC) NeedsProbabilities() bool

NeedsProbabilities method of ROCAUC.

func (ROCAUC) String 

func (rocauc ROCAUC) String() string

String method of ROCAUC.

type Recall 

type Recall struct {
	Class float64
}

Recall measures the fraction of times a true class was predicted.

func (Recall) Apply 

func (recall Recall) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply Recall.

func (Recall) BiggerIsBetter 

func (recall Recall) BiggerIsBetter() bool

BiggerIsBetter method of Recall.

func (Recall) Classification 

func (recall Recall) Classification() bool

Classification method of Recall.

func (Recall) NeedsProbabilities 

func (recall Recall) NeedsProbabilities() bool

NeedsProbabilities method of Recall.

func (Recall) String 

func (recall Recall) String() string

String method of Recall.

type WeightedF1 

type WeightedF1 struct{}

WeightedF1 measures the weighted average F1 score across all classes. This does take class imbalance into account.

func (WeightedF1) Apply 

func (f1 WeightedF1) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply WeightedF1.

func (WeightedF1) BiggerIsBetter 

func (f1 WeightedF1) BiggerIsBetter() bool

BiggerIsBetter method of WeightedF1.

func (WeightedF1) Classification 

func (f1 WeightedF1) Classification() bool

Classification method of WeightedF1.

func (WeightedF1) NeedsProbabilities 

func (f1 WeightedF1) NeedsProbabilities() bool

NeedsProbabilities method of WeightedF1.

func (WeightedF1) String 

func (f1 WeightedF1) String() string

String method of WeightedF1.

type WeightedPrecision 

type WeightedPrecision struct{}

WeightedPrecision measures the weighted average precision across all classes. This does take class imbalance into account.

func (WeightedPrecision) Apply 

func (precision WeightedPrecision) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply WeightedPrecision.

func (WeightedPrecision) BiggerIsBetter 

func (precision WeightedPrecision) BiggerIsBetter() bool

BiggerIsBetter method of WeightedPrecision.

func (WeightedPrecision) Classification 

func (precision WeightedPrecision) Classification() bool

Classification method of WeightedPrecision.

func (WeightedPrecision) NeedsProbabilities 

func (precision WeightedPrecision) NeedsProbabilities() bool

NeedsProbabilities method of WeightedPrecision.

func (WeightedPrecision) String 

func (precision WeightedPrecision) String() string

String method of WeightedPrecision.

type WeightedRecall 

type WeightedRecall struct{}

WeightedRecall measures the weighted average recall across all classes. This does take class imbalance into account.

func (WeightedRecall) Apply 

func (recall WeightedRecall) Apply(yTrue, yPred, weights []float64) (float64, error)

Apply WeightedRecall.

func (WeightedRecall) BiggerIsBetter 

func (recall WeightedRecall) BiggerIsBetter() bool

BiggerIsBetter method of WeightedRecall.

func (WeightedRecall) Classification 

func (recall WeightedRecall) Classification() bool

Classification method of WeightedRecall.

func (WeightedRecall) NeedsProbabilities 

func (recall WeightedRecall) NeedsProbabilities() bool

NeedsProbabilities method of WeightedRecall.

func (WeightedRecall) String 

func (recall WeightedRecall) String() string

String method of WeightedRecall.

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.