# Spectral Path Regression: Directional Chebyshev Harmonics for Interpretable Tabular Learning

Source: [Spectral Path Regression: Directional Chebyshev Harmonics for Interpretable Tabular Learning](<https://www.sota2.com/research/paper/69d48fd67d0172140ec74653>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-08T05:00:04.658Z

Authors: Milo Coombs

Published: 2026-04-05T12:13:59.000Z

## Abstract

Classical approximation bases such as Chebyshev polynomials provide principled and interpretable representations, but their multivariate tensor-product constructions scale exponentially with dimension and impose axis-aligned structure that is poorly matched to real tabular data. We address this by replacing tensorised oscillations with directional harmonic modes of the form $\\cos(\\mathbf{m}^{\\top}\\arccos(\\mathbf{x}))$, which organise multivariate structure by direction in angular space rather than by coordinate index. This representation yields a discrete spectral regression model in which complexity is controlled by selecting a small number of structured frequency vectors (spectral paths), and training reduces to a single closed-form ridge solve with no iterative optimisation. Experiments on standard continuous-feature tabular regression benchmarks show that the resulting models achieve accuracy competitive with strong nonlinear baselines while remaining compact, computationally efficient, and explicitly interpretable through analytic expressions of learned feature interactions.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Tabular Regression on Energy Cooling UCI](<https://www.sota2.com/research/sota/tabular-regression-on-energy-cooling-uci.md>): Tabular Regression; Energy Cooling UCI; Total experiment time (s): 0.4.

- [Tabular Regression on Concrete Slump OpenML](<https://www.sota2.com/research/sota/tabular-regression-on-concrete-slump-openml.md>): Tabular Regression; Concrete Slump OpenML; Total experiment time (s): 0.4.

- [Tabular Regression on Yacht Hydrodynamics OpenML (Total experiment time)](<https://www.sota2.com/research/sota/tabular-regression-on-yacht-hydrodynamics-openml-total-experiment-time.md>): Tabular Regression; Yacht Hydrodynamics OpenML; Total Experiment Time (s): 0.1.

- [Tabular Regression on Cancer Drug Response OpenML](<https://www.sota2.com/research/sota/tabular-regression-on-cancer-drug-response-openml.md>): Tabular Regression; Cancer Drug Response OpenML; Total Time (s): 1.4.

- [Total experiment time (s) on Ankara Weather OpenML](<https://www.sota2.com/research/sota/total-experiment-time-s-on-ankara-weather-openml.md>): Tabular Regression; Ankara Weather OpenML; Total experiment time (s): 0.1.

- [Tabular Regression on Concrete (UCI)](<https://www.sota2.com/research/sota/tabular-regression-on-concrete-uci.md>): Tabular Regression; Concrete (UCI); Total Experiment Time (s): 1.

- [Tabular Regression on Energy Heating (UCI)](<https://www.sota2.com/research/sota/tabular-regression-on-energy-heating-uci.md>): Tabular Regression; Energy Heating (UCI); Total Experiment Time (s): 0.4.

- [Tabular Regression Runtime on Wine Quality (UCI)](<https://www.sota2.com/research/sota/tabular-regression-runtime-on-wine-quality-uci.md>): Tabular Regression; Wine Quality (UCI); Total experiment time (s): 4.4.

- [Tabular Regression on Phishing Websites (UCI)](<https://www.sota2.com/research/sota/tabular-regression-on-phishing-websites-uci.md>): Tabular Regression; Phishing Websites (UCI); Total Experiment Time (s): 5.5.

- [Tabular Regression on Aquatic Toxicity OpenML (Total experiment time (s))](<https://www.sota2.com/research/sota/tabular-regression-on-aquatic-toxicity-openml-total-experiment-time-s.md>): Tabular Regression; Aquatic Toxicity OpenML; Total Experiment Time (s): 0.2.

- [Tabular Regression on Izmir Weather OpenML](<https://www.sota2.com/research/sota/tabular-regression-on-izmir-weather-openml.md>): Tabular Regression; Izmir Weather OpenML; Total Experiment Time (s): 0.2.

- [Tabular Regression on Echocardiogram PMLB](<https://www.sota2.com/research/sota/tabular-regression-on-echocardiogram-pmlb.md>): Tabular Regression; Echocardiogram PMLB; Total Experiment Time (s): 1.5.

- [Tabular Regression on Superconductivity (UCI)](<https://www.sota2.com/research/sota/tabular-regression-on-superconductivity-uci.md>): Tabular Regression; Superconductivity (UCI); Total Experiment Time (s): 6.3.

- [Tabular Regression on Wind Speed PMLB](<https://www.sota2.com/research/sota/tabular-regression-on-wind-speed-pmlb.md>): Tabular Regression; Wind Speed PMLB; Total Experiment Time (s): 7.3.

- [Tabular Regression on CPU Utilisation (PMLB)](<https://www.sota2.com/research/sota/tabular-regression-on-cpu-utilisation-pmlb.md>): Tabular Regression; CPU Utilisation (PMLB); Total experiment time (s): 5.8.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.04091>)

[PDF](<https://arxiv.org/pdf/2604.04091.pdf>)
