Files
elasticsearch-js/docs/reference/api/MlDataframeEvaluationRegressionMetrics.md
2025-05-28 10:02:30 -05:00

9 lines
833 B
Markdown

# `MlDataframeEvaluationRegressionMetrics` [interface-MlDataframeEvaluationRegressionMetrics]
| Name | Type | Description |
| - | - | - |
| `huber` | [MlDataframeEvaluationRegressionMetricsHuber](./MlDataframeEvaluationRegressionMetricsHuber.md) | Pseudo Huber loss function. |
| `mse` | Record<string, any> | Average squared difference between the predicted values and the actual (ground truth) value. For more information, read this wiki article. |
| `msle` | [MlDataframeEvaluationRegressionMetricsMsle](./MlDataframeEvaluationRegressionMetricsMsle.md) | Average squared difference between the logarithm of the predicted values and the logarithm of the actual (ground truth) value. |
| `r_squared` | Record<string, any> | Proportion of the variance in the dependent variable that is predictable from the independent variables. |