Skip to main content

A package to compute the Continuous Ranked Probability Score (CRPS), the Fair-CRPS, and the Adjusted-CRPS. Read the documentation at https://github.com/gouthamnaveen/CRPS

Project description

Documentation

A package to compute the continuous ranked probability score (crps) (Matheson and Winkler, 1976; Hersbach, 2000), the fair-crps (fcrps) (Ferro et al., 2008), and the adjusted-crps (acrps) (Ferro et al., 2008) given an ensemble prediction and an observation.

The CRPS is a negatively oriented score that is used to compare the empirical distribution of an ensemble prediction to a scalar observation.

Read documentation at https://github.com/gouthamnaveen/CRPS

References:

[1] Matheson, J. E. & Winkler, R. L. Scoring Rules for Continuous Probability Distributions. Management Science 22, 1087–1096 (1976).

[2] Hersbach, H. Decomposition of the Continuous Ranked Probability Score for Ensemble Prediction Systems. Wea. Forecasting 15, 559–570 (2000).

[3] Ferro, C. A. T., Richardson, D. S. & Weigel, A. P. On the effect of ensemble size on the discrete and continuous ranked probability scores. Meteorological Applications 15, 19–24 (2008).

Installation:

pip install CRPS

Parameters:

ensemble_members: numpy.ndarray

The predicted ensemble members. They will be sorted in ascending order automatically.

Ex: np.array([2.1,3.5,4.7,1.2,1.3,5.2,5.3,4.2,3.1,1.7])

observation: float

The observed scalar.

Ex: 5.4

adjusted_ensemble_size: int, optional

The size the ensemble needs to be adjusted to before computing the Adjusted Continuous Ranked Probability Score. The default is 200.

Note: The crps becomes equal to acrps when adjusted_ensemble_size equals the length of the ensemble_members.

Method(s):

compute():

Computes the continuous ranked probability score (crps), the fair-crps (fcrps), and the adjusted-crps (acrps).

Returns:

crps,fcrps,acrps

Attributes:

crps: Continuous Ranked Probability Score

It is the integral of the squared difference between the CDF of the forecast ensemble and the observation.

crps

fcrps: Fair-Continuous Ranked Probability Score

It is the crps computed assuming an infinite ensemble size.

fcrps

where m is the current ensemble size (i.e., len(ensemble_members))

acrps: Adjusted-Continuous Ranked Probability Score

It is the crps computed assuming an ensemble size of M.

acrps

where M is the adjusted_ensemble_size

Demonstration:

import numpy as np
import CRPS.CRPS as pscore

Example - 1:

In [1]: pscore(np.arange(1,5),3.5).compute()
Out[1]: (0.625, 0.4166666666666667, 0.42083333333333334)

Example - 2:

In [2]: crps,fcrps,acrps = pscore(np.arange(1,11),8.3,50).compute()
In [3]: crps
Out[3]: 1.6300000000000003
In [4]: fcrps
Out[4]: 1.446666666666667
In [5]: acrps
Out[5]: 1.4833333333333336

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

CRPS-2.0.4.tar.gz (7.7 kB view details)

Uploaded Source

Built Distribution

CRPS-2.0.4-py3-none-any.whl (8.8 kB view details)

Uploaded Python 3

File details

Details for the file CRPS-2.0.4.tar.gz.

File metadata

  • Download URL: CRPS-2.0.4.tar.gz
  • Upload date:
  • Size: 7.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.7.1 importlib_metadata/4.11.3 pkginfo/1.8.2 requests/2.27.1 requests-toolbelt/0.9.1 tqdm/4.64.0 CPython/3.7.13

File hashes

Hashes for CRPS-2.0.4.tar.gz
Algorithm Hash digest
SHA256 f0d24da83e4e5a0bb9e84927f068e525613579001bcccb76274e112f13511229
MD5 0af417d8eadac8a4db7d1eb789254d55
BLAKE2b-256 ad79fa0b68194fb3112fc9320c2f03119d0d9e7d63226800b8617af233107b43

See more details on using hashes here.

Provenance

File details

Details for the file CRPS-2.0.4-py3-none-any.whl.

File metadata

  • Download URL: CRPS-2.0.4-py3-none-any.whl
  • Upload date:
  • Size: 8.8 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.7.1 importlib_metadata/4.11.3 pkginfo/1.8.2 requests/2.27.1 requests-toolbelt/0.9.1 tqdm/4.64.0 CPython/3.7.13

File hashes

Hashes for CRPS-2.0.4-py3-none-any.whl
Algorithm Hash digest
SHA256 2a1fd666702753ff9e1b951bcd958f9c5046eafc023b77df0bdf60539932ba0f
MD5 c9128739b307b901844462024f5f1133
BLAKE2b-256 65c3759461ea6293ec16f6a5a7a273f09225e18fa7b3f6d2c2878486a6ead83e

See more details on using hashes here.

Provenance

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page