Skip to content

RegressionMetrics

A collection of regression metrics.

Stub code in RegressionMetrics.sdsstub 

  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
class RegressionMetrics {
    /**
     * Summarize regression metrics on the given data.
     *
     * @param predicted The predicted target values produced by the regressor.
     * @param expected The expected target values.
     *
     * @result metrics A table containing the regression metrics.
     */
    @Pure
    @Category(DataScienceCategory.ModelEvaluationQMetric)
    static fun summarize(
        predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
        expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
    ) -> metrics: Table

    /**
     * Compute the coefficient of determination (R²) on the given data.
     *
     * The coefficient of determination compares the regressor's predictions to another model that always predicts the
     * mean of the target values. It is a measure of how well the regressor explains the variance in the target values.
     *
     * The **higher** the coefficient of determination, the better the regressor. Results range from negative infinity
     * to 1.0. You can interpret the coefficient of determination as follows:
     *
     * | R²         | Interpretation                                                                             |
     * | ---------- | ------------------------------------------------------------------------------------------ |
     * | 1.0        | The model perfectly predicts the target values. Did you overfit?                           |
     * | (0.0, 1.0) | The model is better than predicting the mean of the target values. You should be here.     |
     * | 0.0        | The model is as good as predicting the mean of the target values. Try something else.      |
     * | (-∞, 0.0)  | The model is worse than predicting the mean of the target values. Something is very wrong. |
     *
     * **Note:** Some other libraries call this metric `r2_score`.
     *
     * @param predicted The predicted target values produced by the regressor.
     * @param expected The expected target values.
     *
     * @result coefficientOfDetermination The calculated coefficient of determination.
     */
    @Pure
    @PythonName("coefficient_of_determination")
    @Category(DataScienceCategory.ModelEvaluationQMetric)
    static fun coefficientOfDetermination(
        predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
        expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
    ) -> coefficientOfDetermination: Float

    /**
     * Compute the mean absolute error (MAE) on the given data.
     *
     * The mean absolute error is the average of the absolute differences between the predicted and expected target
     * values. The **lower** the mean absolute error, the better the regressor. Results range from 0.0 to positive
     * infinity.
     *
     * @param predicted The predicted target values produced by the regressor.
     * @param expected The expected target values.
     *
     * @result meanAbsoluteError The calculated mean absolute error.
     */
    @Pure
    @PythonName("mean_absolute_error")
    @Category(DataScienceCategory.ModelEvaluationQMetric)
    static fun meanAbsoluteError(
        predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
        expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
    ) -> meanAbsoluteError: Float

    /**
     * Compute the mean directional accuracy (MDA) on the given data.
     *
     * This metric compares two consecutive target values and checks if the predicted direction (down/unchanged/up)
     * matches the expected direction. The mean directional accuracy is the proportion of correctly predicted
     * directions. The **higher** the mean directional accuracy, the better the regressor. Results range from 0.0 to
     * 1.0.
     *
     * This metric is useful for time series data, where the order of the target values has a meaning. It is not useful
     * for other types of data. Because of this, it is not included in the `summarize` method.
     *
     * @param predicted The predicted target values produced by the regressor.
     * @param expected The expected target values.
     *
     * @result meanDirectionalAccuracy The calculated mean directional accuracy.
     */
    @Pure
    @PythonName("mean_directional_accuracy")
    @Category(DataScienceCategory.ModelEvaluationQMetric)
    static fun meanDirectionalAccuracy(
        predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
        expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
    ) -> meanDirectionalAccuracy: Float

    /**
     * Compute the mean squared error (MSE) on the given data.
     *
     * The mean squared error is the average of the squared differences between the predicted and expected target
     * values. The **lower** the mean squared error, the better the regressor. Results range from 0.0 to positive
     * infinity.
     *
     * **Note:** To get the root mean squared error (RMSE), take the square root of the result.
     *
     * @param predicted The predicted target values produced by the regressor.
     * @param expected The expected target values.
     *
     * @result meanSquaredError The calculated mean squared error.
     */
    @Pure
    @PythonName("mean_squared_error")
    @Category(DataScienceCategory.ModelEvaluationQMetric)
    static fun meanSquaredError(
        predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
        expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
    ) -> meanSquaredError: Float

    /**
     * Compute the median absolute deviation (MAD) on the given data.
     *
     * The median absolute deviation is the median of the absolute differences between the predicted and expected
     * target values. The **lower** the median absolute deviation, the better the regressor. Results range from 0.0 to
     * positive infinity.
     *
     * @param predicted The predicted target values produced by the regressor.
     * @param expected The expected target values.
     *
     * @result medianAbsoluteDeviation The calculated median absolute deviation.
     */
    @Pure
    @PythonName("median_absolute_deviation")
    @Category(DataScienceCategory.ModelEvaluationQMetric)
    static fun medianAbsoluteDeviation(
        predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
        expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
    ) -> medianAbsoluteDeviation: Float
}

coefficientOfDetermination

Compute the coefficient of determination (R²) on the given data.

The coefficient of determination compares the regressor's predictions to another model that always predicts the mean of the target values. It is a measure of how well the regressor explains the variance in the target values.

The higher the coefficient of determination, the better the regressor. Results range from negative infinity to 1.0. You can interpret the coefficient of determination as follows:

R² Interpretation
1.0 The model perfectly predicts the target values. Did you overfit?
(0.0, 1.0) The model is better than predicting the mean of the target values. You should be here.
0.0 The model is as good as predicting the mean of the target values. Try something else.
(-∞, 0.0) The model is worse than predicting the mean of the target values. Something is very wrong.

Note: Some other libraries call this metric r2_score.

Parameters:

Name Type Description Default
predicted union<Column<Any>, TabularDataset, TimeSeriesDataset> The predicted target values produced by the regressor. -
expected union<Column<Any>, TabularDataset, TimeSeriesDataset> The expected target values. -

Results:

Name Type Description
coefficientOfDetermination Float The calculated coefficient of determination.
Stub code in RegressionMetrics.sdsstub 

48
49
50
51
52
53
54
@Pure
@PythonName("coefficient_of_determination")
@Category(DataScienceCategory.ModelEvaluationQMetric)
static fun coefficientOfDetermination(
    predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
    expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
) -> coefficientOfDetermination: Float

meanAbsoluteError

Compute the mean absolute error (MAE) on the given data.

The mean absolute error is the average of the absolute differences between the predicted and expected target values. The lower the mean absolute error, the better the regressor. Results range from 0.0 to positive infinity.

Parameters:

Name Type Description Default
predicted union<Column<Any>, TabularDataset, TimeSeriesDataset> The predicted target values produced by the regressor. -
expected union<Column<Any>, TabularDataset, TimeSeriesDataset> The expected target values. -

Results:

Name Type Description
meanAbsoluteError Float The calculated mean absolute error.
Stub code in RegressionMetrics.sdsstub 

68
69
70
71
72
73
74
@Pure
@PythonName("mean_absolute_error")
@Category(DataScienceCategory.ModelEvaluationQMetric)
static fun meanAbsoluteError(
    predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
    expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
) -> meanAbsoluteError: Float

meanDirectionalAccuracy

Compute the mean directional accuracy (MDA) on the given data.

This metric compares two consecutive target values and checks if the predicted direction (down/unchanged/up) matches the expected direction. The mean directional accuracy is the proportion of correctly predicted directions. The higher the mean directional accuracy, the better the regressor. Results range from 0.0 to 1.0.

This metric is useful for time series data, where the order of the target values has a meaning. It is not useful for other types of data. Because of this, it is not included in the summarize method.

Parameters:

Name Type Description Default
predicted union<Column<Any>, TabularDataset, TimeSeriesDataset> The predicted target values produced by the regressor. -
expected union<Column<Any>, TabularDataset, TimeSeriesDataset> The expected target values. -

Results:

Name Type Description
meanDirectionalAccuracy Float The calculated mean directional accuracy.
Stub code in RegressionMetrics.sdsstub 

92
93
94
95
96
97
98
@Pure
@PythonName("mean_directional_accuracy")
@Category(DataScienceCategory.ModelEvaluationQMetric)
static fun meanDirectionalAccuracy(
    predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
    expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
) -> meanDirectionalAccuracy: Float

meanSquaredError

Compute the mean squared error (MSE) on the given data.

The mean squared error is the average of the squared differences between the predicted and expected target values. The lower the mean squared error, the better the regressor. Results range from 0.0 to positive infinity.

Note: To get the root mean squared error (RMSE), take the square root of the result.

Parameters:

Name Type Description Default
predicted union<Column<Any>, TabularDataset, TimeSeriesDataset> The predicted target values produced by the regressor. -
expected union<Column<Any>, TabularDataset, TimeSeriesDataset> The expected target values. -

Results:

Name Type Description
meanSquaredError Float The calculated mean squared error.
Stub code in RegressionMetrics.sdsstub 

114
115
116
117
118
119
120
@Pure
@PythonName("mean_squared_error")
@Category(DataScienceCategory.ModelEvaluationQMetric)
static fun meanSquaredError(
    predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
    expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
) -> meanSquaredError: Float

medianAbsoluteDeviation

Compute the median absolute deviation (MAD) on the given data.

The median absolute deviation is the median of the absolute differences between the predicted and expected target values. The lower the median absolute deviation, the better the regressor. Results range from 0.0 to positive infinity.

Parameters:

Name Type Description Default
predicted union<Column<Any>, TabularDataset, TimeSeriesDataset> The predicted target values produced by the regressor. -
expected union<Column<Any>, TabularDataset, TimeSeriesDataset> The expected target values. -

Results:

Name Type Description
medianAbsoluteDeviation Float The calculated median absolute deviation.
Stub code in RegressionMetrics.sdsstub 

134
135
136
137
138
139
140
@Pure
@PythonName("median_absolute_deviation")
@Category(DataScienceCategory.ModelEvaluationQMetric)
static fun medianAbsoluteDeviation(
    predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
    expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
) -> medianAbsoluteDeviation: Float

summarize

Summarize regression metrics on the given data.

Parameters:

Name Type Description Default
predicted union<Column<Any>, TabularDataset, TimeSeriesDataset> The predicted target values produced by the regressor. -
expected union<Column<Any>, TabularDataset, TimeSeriesDataset> The expected target values. -

Results:

Name Type Description
metrics Table A table containing the regression metrics.
Stub code in RegressionMetrics.sdsstub 

18
19
20
21
22
23
@Pure
@Category(DataScienceCategory.ModelEvaluationQMetric)
static fun summarize(
    predicted: union<Column<Any>, TabularDataset, TimeSeriesDataset>,
    expected: union<Column<Any>, TabularDataset, TimeSeriesDataset>
) -> metrics: Table