Search papers, labs, and topics across Lattice.
This study benchmarks the effectiveness of eleven location encoders from the TorchSpatial framework in capturing spatial effects through GeoShapley, a game-theoretic explainer. By evaluating the encoders across three scales鈥攇rid, county, and global鈥攁nd comparing their performance against a raw-coordinate baseline, the authors reveal that while primary coefficient recovery is consistently high, the recovery of secondary coefficients is significantly influenced by scale, particularly at the global level. The findings highlight the nuanced performance of location encoders in spatial modeling, suggesting that raw coordinates may still hold competitive advantages in certain contexts.
Location encoders excel at recovering primary spatial effects, but struggle with secondary effects, especially at larger scales, challenging assumptions about their interpretability.
Location encoders transform geographic coordinates into high dimensional embeddings for downstream machine learning, but it is unclear how well these representations capture interpretable spatial effects. We benchmark whether GeoShapley, a game-theoretic explainer that treats all location features as a single joint player, can recover spatially varying coefficients from models built on location-encoder embeddings. Eleven encoders from the TorchSpatial framework are evaluated against a synthetic process with known coefficients, across three scales (grid, county, global), with and without raw coordinates alongside the embedding, and under untrained and contrastively trained conditions. Measuring recovery as the correlation between estimated and true coefficients, we report how it varies with scale and encoder architecture and compare the embeddings against a raw-coordinate baseline. Recovery of the primary coefficient is consistently high across encoders, whereas recovery of a secondary coefficient is more scale-dependent, differing most at the global scale; the raw-coordinate baseline remains competitive throughout.
