Published October 7, 2022 | Version 2
Dataset Open

Local explanation SHAP approach applied to MIROC5,RCP8.5-forced multi-model ensemble study of GrIS future sea-level contributions

Authors/Creators

  • 1. BRGM

Description

The repository contains materials for analysing the results of the Local explanation named SHAP-CTREE (Redelmeier et al., 2020) approach applied to the MIROC5,RCP8.5-forced multi-model ensemble study of GrIS future sea-level contributions from Goelzer et al. (2020).

The available files are:
- run_SupplMat.R: the main R script to perform the diagnostics and the different analyses (levels 1 - 3)
- utilsPLOT.R: functions for plotting
- Diagnostics.zip: the zip file with the png figures, named 'GrIS_CaseXXX_yYYY.png', that depict the diagnostic for case XXX for prediction time YYY
- SupplementaryMaterials.zip
- RData files for each prediction time YYY "Shapley_yYYY" with:
S: matrix N=55 cases x d+1: SHAP values for the d inputs (+ average sea level value at time YYY)
YHAT: ML-based predictions of the sea level for the 55 cases
YTRUE: true values for the 55 cases
mae: mean absolute error
- RData file containing the design of experiments "DOE_GrIS_MIROC5-RCP85.RData"
doe: matrix with values of the d=9 inputs

These constitute the supplementary materials of Rohmer et al. (2022, The Cryosphere). All technical details are provided in this reference.

Files

Diagnostics.zip

Files (2.7 MB)

Name Size
md5:e2933a971e34c58145953aad4aa6aca7
2.6 MB Preview Download
md5:41db56c3367964244a037148c6a304bb
6.7 kB Download
md5:e92d5122c1ab86eafb56dc009f53e265
95.4 kB Preview Download
md5:4b4a462839b050cc790609a5efdc614c
7.4 kB Download

Additional details

Funding

European Commission
PROTECT - PROjecTing sEa-level rise : from iCe sheets to local implicaTions 869304

References

  • Goelzer, H., Nowicki, S., Payne, A., Larour, E., Seroussi, H., Lipscomb, W. H., et al.: The future sea-level contribution of the Greenland ice sheet: a multi-model ensemble study of ISMIP6, The Cryosphere, 14(9), 3071-3096, 2020.
  • Rohmer, J., Thieblemont, R., Le Cozannet, G., Goelzer, H. Durand, G.: Improving interpretation of sea-level projections through a machine-learning-based local explanation approach, The Cryosphere, submitted.
  • Redelmeier, A., Jullum, M., and Aas, K.: Explaining predictive models with mixed features using Shapley values and conditional inference trees, in International Cross-Domain Conference for Machine Learning and Knowledge Extraction (pp. 117-137). Springer, Cham, 2020.