Log10 (Mean of mean squared displacements @ 600K - Ų) - log_msd_mean_600

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	0.03622	0.09852	0.02664	0.07194	0.9865	0.89944
min	0.0352	0.0938	0.0264	0.0689	0.9858	0.8877
max	0.0369	0.103	0.0269	0.0742	0.9874	0.9098
std	0.000584466	0.00346491	0.000162481	0.0023821	0.000517687	0.00826525

RF - MATMINER+LOBSTER

	train_rmse	test_rmse	train_errors	test_errors	train_r2	test_r2
mean	0.0357	0.09858	0.02612	0.0717	0.98688	0.89952
min	0.0348	0.0962	0.0257	0.0683	0.9865	0.8931
max	0.0361	0.1024	0.0263	0.0741	0.9877	0.9056
std	0.000501996	0.00221124	0.000213542	0.00223517	0.000430813	0.00508936

MODNet - MATMINER

	train_rmse	test_rmse	train_errors	test_errors	train_r2	test_r2
mean	0.0343	0.07474	0.02326	0.05266	0.98764	0.94148
min	0.031	0.0668	0.021	0.0485	0.98	0.9169
max	0.0443	0.0877	0.0304	0.0597	0.99	0.9543
std	0.00505134	0.00732	0.00360311	0.00376436	0.00384947	0.0130464

MODNet - MATMINER+LOBSTER

	train_rmse	test_rmse	train_errors	test_errors	train_r2	test_r2
mean	0.03652	0.07694	0.02464	0.05482	0.98576	0.93834
min	0.0294	0.0673	0.0196	0.0499	0.9756	0.9202
max	0.0484	0.086	0.0317	0.06	0.9912	0.9512
std	0.00690663	0.00684474	0.00435229	0.00384572	0.00563191	0.0110721

Corrected resampled t-test on 10-fold CV

Summary

	t_stat	p_value	significance_stars	d_av	rel_improvement	percent_folds_improved
RF	2.34985	0.0216577	*	0.193262	1.91178	90
MODNet	0.131332	0.449201		0.0228575	0.270577	70

---

Model Explainer

PFI

SHAP

---

SISSO Models

Rung 1

1D descriptor

\[\begin{align*} & log\_msd\_mean\_600 = -1.27 \\ & -6.386\left(\frac{ ElementProperty_{MagpieData\_mean\_NUnfilled} }{ VoronoiFingerprint_{mean\_Voro\_area\_sum} } \right) \end{align*}\]

2D descriptor

\[\begin{align*} & log\_msd\_mean\_600 = -3.573 \\ & + 0.854\left(\exp{ \left(-ElementProperty_{MagpieData\_minimum\_NValence} \right) } \right) \\ & + 0.014\left(VoronoiFingerprint_{mean\_Voro\_area\_sum} \\ + ElementProperty_{MagpieData\_maximum\_MendeleevNumber}\right) \end{align*}\]

Rung 2

1D descriptor

\[\begin{align*} & log\_msd\_all\_600 = -0.74 \\ & -0.158\left(\left(ElementProperty_{MagpieData\_mean\_NUnfilled} + \color{#cc3366}{Loewdin_{std}}\right) \\ + \left(\sqrt[3]{ GaussianSymmFunc_{mean\_G4\_0.005\_1.0\_1.0} }\right)\right) \end{align*}\]

2D descriptor

\[\begin{align*} & log\_msd\_mean\_600 = -0.759 \\ & -0.271\left(\frac{ \left(\left|ElementProperty_{MagpieData\_minimum\_NUnfilled} - {\color{#cc3366}{Loewdin_{std}}}\right|\right) }{ \left(OPSiteFingerprint_{mean\_sgl\_bd\_CN\_1} + ElementProperty_{MagpieData\_mean\_NUnfilled}\right) } \right) \\ & -0.163\left(\left(ElementProperty_{MagpieData\_mean\_NUnfilled} \\ - AtomicPackingEfficiency_{dist\_from\_1\_clusters\_\_APE\_\_less\_0.010}\right) \\ + \left(\sqrt[3]{ GaussianSymmFunc_{mean\_G4\_0.005\_1.0\_1.0} }\right)\right) \end{align*}\]

---

Misc

ARFS n-iter convergence checks

MAE/ fold from 10-fold CV

Alternative visual summary of input data for t-test