Log10 (Bulk modulus - GPa) - log_k_vrh

ARFS Top features

ARFS selected descriptors

---

Correlation analysis

Distance correlation

Dependency graphs

Feature learnability

---

Model performance

5-Fold CV Metrics overview

RF - MATMINER

	train_rmse	test_rmse	train_errors	test_errors	train_r2	test_r2
mean	0.05888	0.1602	0.03082	0.08502	0.98278	0.87066
min	0.0486	0.126	0.0294	0.0773	0.9805	0.7842
max	0.0626	0.2181	0.032	0.093	0.9881	0.9186
std	0.00530863	0.032413	0.000830422	0.00605785	0.00288541	0.0457325

RF - MATMINER+LOBSTER

	train_rmse	test_rmse	train_errors	test_errors	train_r2	test_r2
mean	0.05636	0.14536	0.02896	0.07868	0.9842	0.89198
min	0.0463	0.1111	0.0278	0.0723	0.9815	0.7963
max	0.0608	0.2118	0.0301	0.0844	0.9892	0.9367
std	0.00527886	0.0354314	0.000755248	0.00494668	0.00276767	0.0491209

MODNet - MATMINER

	train_rmse	test_rmse	train_errors	test_errors	train_r2	test_r2
mean	0.07562	0.11798	0.0169	0.05536	0.96896	0.9255
min	0.0264	0.0854	0.0152	0.0494	0.9611	0.824
max	0.0884	0.1969	0.0184	0.0604	0.9965	0.9622
std	0.0246114	0.0413765	0.00105451	0.00483554	0.0137889	0.0516946

MODNet - MATMINER+LOBSTER

	train_rmse	test_rmse	train_errors	test_errors	train_r2	test_r2
mean	0.07764	0.11406	0.01956	0.05336	0.96694	0.92956
min	0.026	0.0792	0.0156	0.0456	0.9491	0.8275
max	0.1004	0.1949	0.0323	0.0583	0.9966	0.9678
std	0.0263257	0.0421391	0.00639206	0.00542977	0.0158044	0.0518521

Corrected resampled t-test on 10-fold CV

Summary

	t_stat	p_value	significance_stars	d_av	rel_improvement	percent_folds_improved
RF	3.95838	0.0016562	**	0.660537	6.99669	100
MODNet	1.18162	0.133819		0.184884	2.40508	80

---

Model Explainer

PFI

SHAP

---

SISSO Models

Rung 1

1D descriptor

\[\begin{align*} & log\_k\_vrh = 3.326 \\ & -0.007\left(ElementProperty_{MagpieData\_maximum\_MendeleevNumber} \\ * AverageBondLength_{mean\_Average\_bond\_length}\right) \end{align*}\]

2D descriptor

\[\begin{align*} & log\_k\_vrh = 4.432 \\ & + 0.663\left(\ln{ MaximumPackingEfficiency_{max\_packing\_efficiency} }\right) \\ & -0.009\left(ElementProperty_{MagpieData\_maximum\_MendeleevNumber} \\ * AverageBondLength_{mean\_Average\_bond\_length}\right) \end{align*}\]

Rung 2

1D descriptor

\[\begin{align*} & log\_k\_vrh = 3.7 \\ & -0.006\left(\frac{ \left(ElementProperty_{MagpieData\_maximum\_MendeleevNumber} * AverageBondLength_{mean\_Average\_bond\_length}\right) }{ \left(\sqrt[3]{ MaximumPackingEfficiency_{max\_packing\_efficiency} }\right) } \right) \end{align*}\]

2D descriptor

\[\begin{align*} & log\_k\_vrh = 2.389 \\ & + 0.099\left(\left(ElementProperty_{MagpieData\_mean\_NUnfilled} + \color{#cc3366}{Mulliken_{std}}\right) \\ - \left(\frac{ AGNIFingerPrint_{std\_dev\_AGNI\_dir\_y\_eta\_2.89e+00} }{ MaximumPackingEfficiency_{max\_packing\_efficiency} } \right)\right) \\ & - 0.079\left(\frac{ \left(AverageBondLength_{mean\_Average\_bond\_length}^2\right) }{ \left(\sqrt{ MaximumPackingEfficiency_{max\_packing\_efficiency} }\right) } \right) \end{align*}\]

---

Misc

ARFS n-iter convergence checks

MAE/ fold from 10-fold CV

Alternative visual summary of input data for t-test