SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.22422 · 2026

Revisiting Chebyshev Polynomial and Anisotropic RBF Models for Tabular Regression

Luciano Gerber, Huw Lloyd

arXiv · PDF · Open in the Atlas

Code that ran

12 of 18 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 2 violated it and 1 has no contract verdict; 8 failures of our instrument; 6 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2602.22422.json · badge: /api/ran/2602.22422.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
gerberl/poly-erbf-benchmark canonical (classifier label)
classifier basis: census
12 ran · 6 no recorded run
FunctionStatusWhere it lives
adjusted_r2 Ran
ran, honoured the contract we drafted
gerberl/poly-erbf-benchmark/perbf/evaluation/metrics.py
code served (permissive licence) · get_code("e519eeaa38f2191e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

ARGS = {
    "r2_score": 0.85,
    "n_samples": 100,
    "n_features": 10,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_ranks_and_summary Ran
ran; our contract draft was wrong, not the code
gerberl/poly-erbf-benchmark/making_plots/generate_rank_linedot_plots_04feb26.py
code served (permissive licence) · get_code("e45f10d15a379cbc")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned DataFrame. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd
import numpy as np

# Create a sample dataframe with multiple models and datasets
np.random.seed(0)

models = ['model_a', 'model_b', 'model_c']
datasets = ['dataset_1', 'dataset_2', 'dataset_3']

# Create data with some missing combinations
data = []
for model in models:
    for dataset in datasets:
        # Skip some combinations to test missing rows handling
        if not (model == 'model_a' and dataset == 'dataset_1'):
            metric_value = np.random.uniform(0, 100)
            data.append({
                'model': model,
                'dataset': dataset,
                'accuracy': metric_value
            })

df = pd.DataFrame(data)

ARGS = {
    'df': df,
    'metric': 'accuracy',
    'lower_is_better': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
dt_factory Ran
ran; our contract draft was wrong, not the code
gerberl/poly-erbf-benchmark/perbf/tuning/optuna_cv.py
code served (permissive licence) · get_code("c1cc1e0408659375")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned DecisionTreeRegressor. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import optuna
from sklearn.tree import DecisionTreeRegressor

# Create an Optuna trial object
# We'll use a simple study and get a trial from it
study = optuna.create_study()
trial = study.ask()

# The function takes only one parameter: trial
# This is an optuna.Trial object that will be used to suggest hyperparameters
ARGS = {
    'trial': trial
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
optuna, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DecisionTreeRegressor — a DecisionTreeRegressor output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DecisionTreeRegressor output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_default_params Ran
ran; our contract draft was wrong, not the code
gerberl/poly-erbf-benchmark/perbf/defaults.py
code served (permissive licence) · get_code("6c98e0c96c494266")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# Test with a valid model name
model_name_valid = 'ridge'

ARGS = {
    'model_name': model_name_valid,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_eval_cv Ran
ran; our contract draft was wrong, not the code
gerberl/poly-erbf-benchmark/perbf/evaluation/cv.py
code served (permissive licence) · get_code("123ff938e3796f00")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned RepeatedKFold. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

from sklearn.model_selection import RepeatedKFold

ARGS = {
    "n_splits": 4,
    "n_repeats": 5,
    "random_state": 42,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type RepeatedKFold — a RepeatedKFold output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a RepeatedKFold output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
linedotplot Ran
ran on a synthesized input
gerberl/poly-erbf-benchmark/making_plots/generate_rank_linedot_plots_04feb26.py
code served (permissive licence) · get_code("46489d599ed1d889")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# Create sample data for linedotplot
np.random.seed(0)

# Create a DataFrame with the required columns
data = pd.DataFrame({
    'group': ['A', 'B', 'C', 'D'],
    'score': [0.85, 0.72, 0.91, 0.68],
    'error': [0.05, 0.03, 0.07, 0.04],
    'error_low': [0.80, 0.69, 0.84, 0.64],
    'error_high': [0.90, 0.75, 0.98, 0.72]
})

ARGS = {
    'data': data,
    'x': 'score',
    'y': 'group',
    'err': 'error',
    'err_low': 'error_low',
    'err_high': 'error_high',
    'xlabel': 'Performance Score',
    'ylabel': 'Groups',
    'title': 'Ranking with Error Bars',
    'xlim': None,
    'xlim_pad': 0.3,
    'width': 4.5,
    'height_per_row': 0.45,
    'dot_size': 7,
    'dot_linewidth': 0.5,
    'dot_edgecolor': 'w',
    'palette': 'flare',
    'annotate': True,
    'annot_fmt': '.2f',
    'annot_offset': 0.35,
    'annot_size': 7,
    'x_tick_spacing': None,
    'ax': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, pandas, seaborn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
llfrr Ran
ran; our contract draft was wrong, not the code
gerberl/poly-erbf-benchmark/perbf/analysis/discontinuity_smoothness.py
code served (permissive licence) · get_code("f86f522ffa020d53")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seed for reproducibility
np.random.seed(42)

# Create small synthetic dataset
# X: 8 samples, 3 features
X = np.random.randn(8, 3).astype(np.float64)

# y: 8 target values
y = np.random.randn(8).astype(np.float64)

# Define ARGS dictionary with all parameters
ARGS = {
    'X': X,
    'y': y,
    'k': 5,
    'sample_size': None,
    'random_state': 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
llqd_distribution Ran
ran; our contract draft was wrong, not the code
gerberl/poly-erbf-benchmark/perbf/analysis/discontinuity_smoothness.py
code served (permissive licence) · get_code("fc36bbb563953726")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
from sklearn.neighbors import NearestNeighbors

np.random.seed(42)

# Create small synthetic dataset
# X: 8 samples, 3 features
X = np.random.randn(8, 3).astype(np.float32)

# y: 8 target values
y = np.random.randn(8).astype(np.float32)

# Define all parameters
ARGS = {
    'X': X,
    'y': y,
    'k': 3,
    'sample_size': None,
    'random_state': 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
mae Ran
ran, violated the contract we drafted
gerberl/poly-erbf-benchmark/perbf/evaluation/metrics.py
code served (permissive licence) · get_code("c859d072a2ab4fd7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seeds for reproducibility
np.random.seed(0)

# Create small numpy arrays for y_true and y_pred
# These should be 1D arrays of the same shape
y_true = np.array([1.0, 2.0, 3.0, 4.0, 5.0])
y_pred = np.array([1.1, 2.2, 2.9, 4.1, 4.8])

ARGS = {
    "y_true": y_true,
    "y_pred": y_pred,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
ntj_stats Ran
ran; our contract draft was wrong, not the code
gerberl/poly-erbf-benchmark/perbf/analysis/discontinuity_smoothness.py
code served (permissive licence) · get_code("5a46c2913eaef735")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seed for reproducibility
np.random.seed(42)

# Create small synthetic dataset
# X: feature matrix with 10 samples and 4 features
X = np.random.randn(10, 4)

# y: target values with some variation to get meaningful NTJ statistics
y = np.random.randn(10) * 5 + np.arange(10)

# Define all arguments for ntj_stats
ARGS = {
    'X': X,
    'y': y,
    'k': 5,
    'sample_size': None,
    'random_state': 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
preprocess_fold Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
gerberl/poly-erbf-benchmark/perbf/preprocessing/fold_preprocessor.py
code served (permissive licence) · get_code("e896be1606d3e09c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import pandas as pd

# Create small synthetic data for testing
np.random.seed(0)

# Training data: 5 samples, 4 features (3 numeric, 1 categorical)
X_train = np.array([
    [1.0, 2.0, 3.0, 'a'],
    [2.0, 3.0, 4.0, 'b'],
    [3.0, 4.0, 5.0, 'a'],
    [4.0, 5.0, 6.0, 'c'],
    [5.0, 6.0, 7.0, 'b'],
], dtype=object)

# Convert to proper types for numeric and categorical columns
X_train_df = pd.DataFrame({
    'f0': [1.0, 2.0, 3.0, 4.0, 5.0],
    'f1': [2.0, 3.0, 4.0, 5.0, 6.0],
    'f2': [3.0, 4.0, 5.0, 6.0, 7.0],
    'cat': ['a', 'b', 'a', 'c', 'b']
})
X_train = X_train_df.values

y_train = np.array([10.0, 20.0, 15.0, 25.0, 30.0])

# Validation data: 2 samples, 4 features
X_val = np.array([
    [2.5, 3.5, 4.5, 'a'],
    [3.5, 4.5, 5.5, 'b'],
], dtype=object)

X_val_df = pd.DataFrame({
    'f0': [2.5, 3.5],
    'f1': [3.5, 4.5],
    'f2': [4.5, 5.5],
    'cat': ['a', 'b']
})
X_val = X_val_df.values

ARGS = {
    'X_train': X_train,
    'y_train': y_train,
    'X_val': X_val,
    'scale': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
rmse Ran
ran, violated the contract we drafted
gerberl/poly-erbf-benchmark/perbf/evaluation/metrics.py
code served (permissive licence) · get_code("4460afc8e6239c5b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Set random seed for reproducibility
np.random.seed(0)

# Create small numpy arrays for y_true and y_pred
y_true = np.array([1.0, 2.0, 3.0, 4.0, 5.0])
y_pred = np.array([1.1, 2.2, 2.9, 4.1, 4.8])

ARGS = {
    "y_true": y_true,
    "y_pred": y_pred,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
extract_model_info Not yet run
unverified (harvested, no recorded run)
gerberl/poly-erbf-benchmark/perbf/analysis/model_complexity.py
code served (permissive licence) · get_code("a512e8881320a33b")
How it ran

no invocation recorded.

make_rank_plot Not yet run
unverified (harvested, no recorded run)
gerberl/poly-erbf-benchmark/making_plots/generate_rank_linedot_plots_04feb26.py
code served (permissive licence) · get_code("c1bfc95ed2e8eafc")
How it ran

no invocation recorded.

quick_smoothness_check Not yet run
unverified (harvested, no recorded run)
gerberl/poly-erbf-benchmark/perbf/analysis/model_recommender.py
code served (permissive licence) · get_code("b19c6838d02bb2ac")
How it ran

no invocation recorded.

recommend_model_family Not yet run
unverified (harvested, no recorded run)
gerberl/poly-erbf-benchmark/perbf/analysis/model_recommender.py
code served (permissive licence) · get_code("308b4e7ee8d568b8")
How it ran

no invocation recorded.

rf_factory Not yet run
unverified (harvested, no recorded run)
gerberl/poly-erbf-benchmark/perbf/tuning/optuna_cv.py
code served (permissive licence) · get_code("384ccbca8697cf19")
How it ran

no invocation recorded.

ridge_factory Not yet run
unverified (harvested, no recorded run)
gerberl/poly-erbf-benchmark/perbf/tuning/optuna_cv.py
code served (permissive licence) · get_code("cd0a6786665a81d1")
How it ran

no invocation recorded.

Also present in other papers' code

Of 18 samples lifted from this paper's linked repositories, 18 carry a function identity and 0 do not. Of the 18 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 18 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Smooth-basis models such as Chebyshev polynomial regressors and radial basis function (RBF) networks are well established in numerical analysis. Their continuously differentiable prediction surfaces suit surrogate optimisation, sensitivity analysis, and other settings where the response varies gradually with inputs. Despite these properties, smooth models seldom appear in tabular regression, where tree ensembles dominate. We ask whether they can compete, benchmarking models across 55 regression datasets organised by application domain. We develop an anisotropic RBF network with data-driven centre placement and gradientbased width optimisation, a ridge-regularised Chebyshev polynomial regressor, and a smooth-tree hybrid (Chebyshev model tree); all three are released as scikit-learn-compatible packages. We benchmark these against tree ensembles, a pre-trained transformer, and standard baselines, evaluating accuracy alongside generalisation behaviour. The transformer ranks first on accuracy across a majority of datasets, but its GPU dependence, inference latency, and dataset-size limits constrain deployment in the CPU-based settings common across applied science and industry. Among CPU-viable models, smooth models and tree ensembles are statistically tied on accuracy, but the former tend to exhibit tighter generalisation gaps. We recommend routinely including smooth-basis models in the candidate pool, particularly when downstream use benefits from tighter generalisation and gradually varying predictions.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2602.22422")
get_code_for_paper("2602.22422")
have("2602.22422")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2602.22422.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2602.22422.svg)](https://syntology.ai/paper/2602.22422)

Connect an agent — have() is free.