Metrics Module¶
- Classes:
- Metric - Abstract Base Class from which all metrics must inherit.
-
class
metrics.Bias¶ Calculate the bias between a reference and target dataset.
-
run(ref_dataset, target_dataset)¶ Calculate the bias between a reference and target dataset.
Note
Overrides BinaryMetric.run()
Parameters: - ref_dataset (
dataset.Dataset) – The reference dataset to use in this metric run. - target_dataset (
dataset.Dataset) – The target dataset to evaluate against the reference dataset in this metric run.
Returns: The difference between the reference and target datasets.
Return type: - ref_dataset (
-
-
class
metrics.BinaryMetric¶ Abstract Base Class from which all binary metrics inherit.
-
run(ref_dataset, target_dataset)¶ Run the metric for the given reference and target datasets.
Parameters: - ref_dataset (
dataset.Dataset) – The Dataset to use as the reference dataset when running the evaluation. - target_dataset (
dataset.Dataset) – The Dataset to use as the target dataset when running the evaluation.
Returns: The result of evaluation the metric on the reference and target dataset.
- ref_dataset (
-
-
class
metrics.Metric¶ Base Metric Class
-
class
metrics.PatternCorrelation¶ Calculate the correlation coefficient between two datasets
-
run(ref_dataset, target_dataset)¶ Calculate the correlation coefficient between two dataset.
Note
Overrides BinaryMetric.run()
Parameters: - ref_dataset (
dataset.Dataset) – The reference dataset to use in this metric run. - target_dataset (
dataset.Dataset) – The target dataset to evaluate against the reference dataset in this metric run.
Returns: The correlation coefficient between a reference and target dataset.
- ref_dataset (
-
-
class
metrics.RMSError¶ Calculate the Root Mean Square Difference (RMS Error), with the mean calculated over time and space.
-
run(reference_dataset, target_dataset)¶ - Calculate the Root Mean Square Difference (RMS Error), with the mean
- calculated over time and space.
Note
Overrides BinaryMetric.run()
Parameters: - reference_dataset (
dataset.Dataset) – The reference dataset to use in this metric run - target_dataset (
dataset.Dataset) – The target dataset to evaluate against the reference dataset in this metric run
Returns: The RMS error, with the mean calculated over time and space
-
-
class
metrics.SpatialPatternTaylorDiagram¶ Calculate the target to reference ratio of spatial standard deviation and pattern correlation
-
run(ref_dataset, target_dataset)¶ Calculate two metrics to plot a Taylor diagram to compare spatial patterns
Note
Overrides BinaryMetric.run()
Parameters: - ref_dataset (
dataset.Dataset) – The reference dataset to use in this metric run. - target_dataset (
dataset.Dataset) – The target dataset to evaluate against the reference dataset in this metric run.
Returns: standard deviation ratio, pattern correlation coefficient
Return type: :float:’float’,’float’
- ref_dataset (
-
-
class
metrics.StdDevRatio¶ Calculate the standard deviation ratio between two datasets.
-
run(ref_dataset, target_dataset)¶ Calculate the standard deviation ratio.
Note
Overrides BinaryMetric.run()
Parameters: - ref_dataset (
dataset.Dataset) – The reference dataset to use in this metric run. - target_dataset (
dataset.Dataset) – The target dataset to evaluate against the reference dataset in this metric run.
Returns: The standard deviation ratio of the reference and target
- ref_dataset (
-
-
class
metrics.TemporalCorrelation¶ Calculate the temporal correlation coefficients and associated confidence levels between two datasets, using Pearson’s correlation.
-
run(reference_dataset, target_dataset)¶ - Calculate the temporal correlation coefficients and associated
- confidence levels between two datasets, using Pearson’s correlation.
Note
Overrides BinaryMetric.run()
Parameters: - reference_dataset (
dataset.Dataset) – The reference dataset to use in this metric run - target_dataset (
dataset.Dataset) – The target dataset to evaluate against the reference dataset in this metric run
Returns: A 2D array of temporal correlation coefficients and a 2D array of confidence levels associated with the temporal correlation coefficients
-
-
class
metrics.TemporalMeanBias¶ Calculate the bias averaged over time.
-
run(ref_dataset, target_dataset)¶ Calculate the bias averaged over time.
Note
Overrides BinaryMetric.run()
Parameters: - ref_dataset (
dataset.Dataset) – The reference dataset to use in this metric run. - target_dataset (
dataset.Dataset) – The target dataset to evaluate against the reference dataset in this metric run.
Returns: The mean bias between a reference and target dataset over time.
- ref_dataset (
-
-
class
metrics.TemporalStdDev¶ Calculate the standard deviation over the time.
-
run(target_dataset)¶ Calculate the temporal std. dev. for a datasets.
Note
Overrides UnaryMetric.run()
Parameters: target_dataset ( dataset.Dataset) – The target_dataset on which to calculate the temporal standard deviation.Returns: The temporal standard deviation of the target dataset Return type: ndarray
-
-
class
metrics.UnaryMetric¶ Abstract Base Class from which all unary metrics inherit.
-
run(target_dataset)¶ Run the metric for a given target dataset.
Parameters: target_dataset ( dataset.Dataset) – The dataset on which the current metric will be run.Returns: The result of evaluating the metric on the target_dataset.
-
-
metrics.calc_bias(target_array, reference_array, average_over_time=False)¶ Calculate difference between two arrays
Parameters: - target_array (:class:'numpy.ma.core.MaskedArray') – an array to be evaluated, as model output
- reference_array (:class:'numpy.ma.core.MaskedArray') – an array of reference dataset
- average_over_time ('bool') – if True, calculated bias is averaged for the axis=0
Returns: Biases array of the target dataset
Return type: :class:’numpy.ma.core.MaskedArray’
-
metrics.calc_correlation(target_array, reference_array)¶ Calculate the correlation coefficient between two arrays.
Parameters: - target_array (:class:'numpy.ma.core.MaskedArray') – an array to be evaluated, as model output
- reference_array (:class:'numpy.ma.core.MaskedArray') – an array of reference dataset
Returns: pearson’s correlation coefficient between the two input arrays
Return type: :class:’numpy.ma.core.MaskedArray’
-
metrics.calc_histogram_overlap(hist1, hist2)¶ from Lee et al. (2014) :param hist1: a histogram array :type hist1: :class:’numpy.ndarray’ :param hist2: a histogram array with the same size as hist1 :type hist2: :class:’numpy.ndarray’
-
metrics.calc_joint_histogram(data_array1, data_array2, bins_for_data1, bins_for_data2)¶ Calculate a joint histogram of two variables in data_array1 and data_array2 :param data_array1: the first variable :type data_array1: :class:’numpy.ma.core.MaskedArray’ :param data_array2: the second variable :type data_array2: :class:’numpy.ma.core.MaskedArray’ :param bins_for_data1: histogram bin edges for data_array1 :type bins_for_data1: :class:’numpy.ndarray’ :param bins_for_data2: histogram bin edges for data_array2 :type bins_for_data2: :class:’numpy.ndarray’
-
metrics.calc_rmse(target_array, reference_array)¶ Calculate ratio of standard deivations of the two arrays
Parameters: - target_array (:class:'numpy.ma.core.MaskedArray') – an array to be evaluated, as model output
- reference_array (:class:'numpy.ma.core.MaskedArray') – an array of reference dataset
- average_over_time ('bool') – if True, calculated bias is averaged for the axis=0
Returns: root mean square error
Return type: :class:’float’
-
metrics.calc_stddev(array, axis=None)¶ Calculate a sample standard deviation of an array along the array
Parameters: - array (:class:'numpy.ma.core.MaskedArray') – an array to calculate sample standard deviation
- axis ('int') – Axis along which the sample standard deviation is computed.
Returns: sample standard deviation of array
Return type: :class:’numpy.ma.core.MaskedArray’
-
metrics.calc_stddev_ratio(target_array, reference_array)¶ Calculate ratio of standard deivations of the two arrays
Parameters: - target_array (:class:'numpy.ma.core.MaskedArray') – an array to be evaluated, as model output
- reference_array (:class:'numpy.ma.core.MaskedArray') – an array of reference dataset
- average_over_time ('bool') – if True, calculated bias is averaged for the axis=0
Returns: (standard deviation of target_array)/(standard deviation of reference array)
Return type: :class:’float’
-
metrics.wet_spell_analysis(reference_array, threshold=0.1, nyear=1, dt=3.0)¶ Characterize wet spells using sub-daily (hourly) data
Parameters: - reference_array (:class:'numpy.ma.core.MaskedArray') – an array to be analyzed
- threshold ('float') – the minimum amount of rainfall [mm/hour]
- nyear ('int') – the number of discontinous periods
- dt ('float') – the temporal resolution of reference_array
