Improving the interpretability of species distribution models by using local approximations
Angelov, B.
Show abstract
Species Distribution Models (SDMs) are used to generate maps of realised and potential ecological niches for a given species. As any other machine learning technique they can be seen as "black boxes", due to a lack of interpretability. Advances in other areas of applied machine learning can be applied to remedy this problem. In this study we test a new tool relying on Local Interpretable Model-agnostic Explanations (LIME) by comparing its results of other known methods and ecological interpretations from domain experts. The findings confirm that LIME provides consistent and ecologically sound explanations of climate feature importance during the training of SDMs, and that the sdmexplain R package can be used with confidence.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Modelling the effects of bioclimatic characteristics and climate change on the potential distribution of a monospecific species Colophospermum mopane (Benth.) Leonard in southern Africa. 94%
- Transferability of stream benthic macroinvertebrate distribution models to drought-related conditions 93%
- FishPhyloMaker: An R package to generate phylogenies for ray-finned fishes 93%
Similar papers in this journal
Similar papers in this journal
- PointedSDMs -- an R package to help facilitate the construction of integrated species distribution models 95%
- rasterdiv - an Information Theory tailored R package for measuring ecosystem heterogeneity from space: to the origin and back 94%
- embarcadero: Species distribution modelling with Bayesian additive regression trees in R 94%
Similar papers in this journal
Similar papers in this journal
- Predicting species distributions in the open ocean with convolutional neural networks. 95%
- Integrating biodiversity assessments into local conservation planning: the importance of assessing suitable data sources 94%
- Data stochasticity and model parametrisation impact the performance of species distribution models: insights from a simulation study 94%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.