Regression metrics for evaluating model performance on continuous prediction tasks.
meanSquaredError
Mean Squared Error (MSE) - average of squared differences between predictions and actual values.
import { meanSquaredError } from 'scikitjs';
function meanSquaredError(
yTrue: number[] | number[][],
yPred: number[] | number[][],
options?: RegressionMetricOptions
): number | number[];
yTrue
number[] | number[][]
required
True target values (vector or matrix for multi-output)
yPred
number[] | number[][]
required
Predicted target values (must match yTrue dimensions)
Sample weights for weighted average
options.multioutput
'uniform_average' | 'raw_values' | 'variance_weighted'
Defines aggregating strategy for multi-output:
'uniform_average' (default): uniformly weighted mean'raw_values': returns full set of scores'variance_weighted': scores weighted by target variance
Example
import { meanSquaredError } from 'scikitjs';
const yTrue = [3, -0.5, 2, 7];
const yPred = [2.5, 0.0, 2, 8];
const mse = meanSquaredError(yTrue, yPred);
console.log(mse); // 0.375
// Multi-output
const yTrueMulti = [[0.5, 1], [-1, 1], [7, -6]];
const yPredMulti = [[0, 2], [-1, 2], [8, -5]];
const mseMulti = meanSquaredError(yTrueMulti, yPredMulti);
console.log(mseMulti); // 0.708
// Raw values per output
const mseRaw = meanSquaredError(yTrueMulti, yPredMulti, {
multioutput: 'raw_values'
});
console.log(mseRaw); // [0.416, 1.0]
meanAbsoluteError
Mean Absolute Error (MAE) - average of absolute differences between predictions and actual values.
function meanAbsoluteError(
yTrue: number[] | number[][],
yPred: number[] | number[][],
options?: RegressionMetricOptions
): number | number[];
yTrue
number[] | number[][]
required
True target values
yPred
number[] | number[][]
required
Predicted target values
options.multioutput
'uniform_average' | 'raw_values' | 'variance_weighted'
Aggregation strategy for multi-output
Example
const yTrue = [3, -0.5, 2, 7];
const yPred = [2.5, 0.0, 2, 8];
const mae = meanAbsoluteError(yTrue, yPred);
console.log(mae); // 0.5
r2Score
R² (coefficient of determination) - proportion of variance in the dependent variable that is predictable.
function r2Score(
yTrue: number[] | number[][],
yPred: number[] | number[][],
options?: RegressionMetricOptions
): number | number[];
yTrue
number[] | number[][]
required
True target values
yPred
number[] | number[][]
required
Predicted target values
options.multioutput
'uniform_average' | 'raw_values' | 'variance_weighted'
Aggregation strategy for multi-output
Example
const yTrue = [3, -0.5, 2, 7];
const yPred = [2.5, 0.0, 2, 8];
const r2 = r2Score(yTrue, yPred);
console.log(r2); // 0.948
meanAbsolutePercentageError
Mean Absolute Percentage Error (MAPE) - average of absolute percentage errors.
function meanAbsolutePercentageError(
yTrue: number[] | number[][],
yPred: number[] | number[][],
options?: RegressionMetricOptions
): number | number[];
yTrue
number[] | number[][]
required
True target values
yPred
number[] | number[][]
required
Predicted target values
options.multioutput
'uniform_average' | 'raw_values' | 'variance_weighted'
Aggregation strategy for multi-output
Example
const yTrue = [3, -0.5, 2, 7];
const yPred = [2.5, 0.0, 2, 8];
const mape = meanAbsolutePercentageError(yTrue, yPred);
console.log(mape);
explainedVarianceScore
Explained variance score - measures the proportion to which a model accounts for the variation of a dataset.
function explainedVarianceScore(
yTrue: number[] | number[][],
yPred: number[] | number[][],
options?: RegressionMetricOptions
): number | number[];
yTrue
number[] | number[][]
required
True target values
yPred
number[] | number[][]
required
Predicted target values
options.multioutput
'uniform_average' | 'raw_values' | 'variance_weighted'
Aggregation strategy for multi-output
Example
const yTrue = [3, -0.5, 2, 7];
const yPred = [2.5, 0.0, 2, 8];
const evs = explainedVarianceScore(yTrue, yPred);
console.log(evs); // 0.957