Unmasking the physical information inherent to interstellar spectral line profiles with machine learning - I. Application of LTE to HCN and HNC transitions

Home

All issues

Volume 698 (May 2025)

A&A, 698 (2025) A286

Full HTML

Open Access

Table A.1

Performance metrics and temperature predictions for Orion KL hot core from the RF, XGB, and MLP models.

RF
Spectral lines	${\bar{R}}_{t r a i n}^{2}$ $\bar{R}_{train}^{2}$	${\bar{R}}_{t e s t}^{2}$ $\bar{R}_{test}^{2}$	T̄_pred (K)	Std_pred (K)	depth
HCN 5–4	0.99	0.99	90	0	20
HNC 4–3	0.62	0.61	100	2	5
HCN 5–4 HNC 4–3	0.999	0.998	90	1	20
XGB
Spectral lines	${\bar{R}}_{t r a i n}^{2}$ $\bar{R}_{train}^{2}$	${\bar{R}}_{t e s t}^{2}$ $\bar{R}_{test}^{2}$	T̄_pred (K)	Std_pred (K)	depth	learning rate
HCN 5–4	0.99	0.99	90	1	10	0.01
HNC 4–3	0.60	0.58	91	1	5	0.003
HCN 5–4 HNC 4–3	0.999	0.998	89	1	10	0.1
MLP
Spectral lines	${\bar{R}}_{t r a i n}^{2}$ $\bar{R}_{train}^{2}$	${\bar{R}}_{t e s t}^{2}$ $\bar{R}_{test}^{2}$	T̄_pred (K)	Std_pred (K)
HCN 5–4	0.96	0.96	93	3
HNC 4–3	0.62	0.60	105	6
HCN 5–4 HNC 4–3	0.99	0.99	68	8

Notes. The mean coefficients of determination are given by ${\bar{R}}_{t r a i n}^{2}$ $\bar{R}_{{train}}^{2}$ and ${\bar{R}}_{t e s t}^{2}$ $\bar{R}_{{test}}^{2}$ for the RF, XGB, MLP models trained with different combinations of spectral lines as well as the mean values and standard deviations of the temperature prediction for Orion KL hot core source. The adjusted relevant hyperparameters of RF and XGB are also reported. The hyperparameters of MLP are described in Sect. 2.2.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.