lightgbm.EvalResult

class lightgbm.EvalResult(dataset_name, metric_name, metric_value, maximize, metric_std_dev=None)[source]

Bases: NamedTuple

Result from computing an evaluation metric on a dataset.

In lightgbm<4.7.0, evaluation results were stored in tuples like this:

train(): (dataset_name, metric_name, metric_value, maximize)

cv(): (dataset_name, metric_name, mean(metric_value), maximize, std_dev(metric_value))

Parameters:

dataset_name (str) – Unique identifier for the dataset this result was computed on.
metric_name (str) – Unique identifier for the metric (e.g. “rmse”).
metric_value (float) – Value of the evaluation metric.
maximize (bool) – Are higher values better? e.g. True for AUC and False for binary error.
metric_std_dev (float or None) – If not None, the standard deviation of metric values computed over a range of results. For example, used when aggregating over cross-validation folds in cv().

__init__()

Methods

`__init__`()
`count`(value, /)	Return number of occurrences of value.
`index`(value[, start, stop])	Return first index of value.
`is_cv_result`()	Whether the result was created by `cv()`.

Attributes

`dataset_name`	Alias for field number 0
`maximize`	Alias for field number 3
`metric_name`	Alias for field number 1
`metric_std_dev`	Alias for field number 4
`metric_value`	Alias for field number 2

count(value, /): Return number of occurrences of value.

dataset_name: Alias for field number 0

index(value, start=0, stop=9223372036854775807, /)

Return first index of value.

Raises ValueError if the value is not present.

is_cv_result()[source]

Whether the result was created by cv().

If True:

metric_value = mean of metric_name over CV folds

metric_std_dev = standard deviation of metric_name over CV folds

maximize: Alias for field number 3

metric_name: Alias for field number 1

metric_std_dev: Alias for field number 4

metric_value: Alias for field number 2