MetricSet¶
- class lsst.verify.MetricSet(metrics=None)¶
Bases:
JsonSerializationMixin
A collection of
Metric
s.- Parameters:
Attributes Summary
A JSON-serializable object (
list
).Methods Summary
deserialize
([metrics])Deserialize metric JSON objects into a MetricSet instance.
insert
(metric)Insert a
Metric
into the set.items
()Iterate over
(name, metric)
pairs in the set.jsonify_dict
(d)Recursively build JSON-renderable objects on all values in a dict.
keys
()Get a list of metric names included in the set
load_metrics_package
([package_name_or_path, ...])Create a MetricSet from a Verification Framework metrics package.
load_single_package
(metrics_yaml_path)Create a MetricSet from a single YAML file containing metric definitions for a single package.
subset
([package, tags])Create a new
MetricSet
with metrics belonging to a single package and/or tag.update
(other)Merge another
MetricSet
into this one.write_json
(filepath)Write JSON to a file.
Attributes Documentation
Methods Documentation
- classmethod deserialize(metrics=None)¶
Deserialize metric JSON objects into a MetricSet instance.
- Parameters:
- metrics
list
List of metric JSON serializations (typically created by
MetricSet.json
).
- metrics
- Returns:
- insert(metric)¶
Insert a
Metric
into the set.Any pre-existing metric with the same name is replaced
- Parameters:
- metric
Metric
A metric.
- metric
- items()¶
Iterate over
(name, metric)
pairs in the set.
- static jsonify_dict(d)¶
Recursively build JSON-renderable objects on all values in a dict.
- Parameters:
- d
dict
Dictionary to convert into a JSON-serializable object. Values are recursively JSON-ified.
- d
- Returns:
- json_dict
dict
Dictionary that can be serialized to JSON.
- json_dict
Examples
Subclasses can use this method to prepare output in their
json
-method implementation. For example:def json(self): return JsonSerializationMixin.jsonify_dict({ 'value': self.value, })
- keys()¶
Get a list of metric names included in the set
- classmethod load_metrics_package(package_name_or_path='verify_metrics', subset=None)¶
Create a MetricSet from a Verification Framework metrics package.
- Parameters:
- package_name_or_path
str
, optional Name of an EUPS package that hosts metric and specification definition YAML files or the file path to a metrics package.
'verify_metrics'
is the default package, and is where metrics and specifications are defined for most packages.- subset
str
, optional If set, only metrics for this package are loaded. For example, if
subset='validate_drp'
, onlyvalidate_drp
metrics are included in theMetricSet
. This argument is equivalent to theMetricSet.subset
method. Default isNone
.
- package_name_or_path
- Returns:
Notes
EUPS packages that host metrics and specification definitions for the Verification Framework have top-level directories named
'metrics'
and'specs'
. The metrics package chosen with thepackage_name_or_path
argument. The default metric package for LSST Science Pipelines isverify_metrics
.To make a
MetricSet
from a single package’s YAML metric definition file that is not contained in a metrics package, useload_single_package
instead.
- classmethod load_single_package(metrics_yaml_path)¶
Create a MetricSet from a single YAML file containing metric definitions for a single package.
Notes
The YAML file’s name, without extension, is taken as the package name for all metrics.
For example,
validate_drp.yaml
contains metrics that are identified as belonging to thevalidate_drp
package.
- subset(package=None, tags=None)¶
Create a new
MetricSet
with metrics belonging to a single package and/or tag.- Parameters:
- package
str
orlsst.verify.Name
, optional Name of the package to subset metrics by. If the package name is
'pkg_a'
, then metric'pkg_a.metric_1'
would be included in the subset, while'pkg_b.metric_2'
would be excluded.- tagssequence of
str
, optional Tags to select metrics by. These tags must be a subset (
<=
) of theMetric.tags
for the metric to be selected.
- package
- Returns:
- metric_subset
MetricSet
Subset of this metric set containing only metrics belonging to the specified package and/or tag.
- metric_subset
Notes
If both
package
andtag
are provided then the resultingMetricSet
contains the intersection of the package-based and tag-based selections. That is, metrics will belong topackage
and posess the tagtag
.