The S-PLUS: a star/galaxy classification based on a Machine Learning approach. (arXiv:1909.08626v1 [astro-ph.GA])
<a href="http://arxiv.org/find/astro-ph/1/au:+Costa_Duarte_M/0/1/0/all/0/1">M. V. Costa-Duarte</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Sampedro_L/0/1/0/all/0/1">L. Sampedro</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Molino_A/0/1/0/all/0/1">A. Molino</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Xavier_H/0/1/0/all/0/1">H. S. Xavier</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Herpich_F/0/1/0/all/0/1">F. R. Herpich</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Chies_Santos_A/0/1/0/all/0/1">A. L. Chies-Santos</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Barbosa_C/0/1/0/all/0/1">C. E. Barbosa</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Cortesi_A/0/1/0/all/0/1">A. Cortesi</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Schoenell_W/0/1/0/all/0/1">W. Schoenell</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Kanaan_A/0/1/0/all/0/1">A. Kanaan</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Ribeiro_T/0/1/0/all/0/1">T. Ribeiro</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Oliveira_C/0/1/0/all/0/1">C. Mendes de Oliveira</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Akras_S/0/1/0/all/0/1">S. Akras</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Alvarez_Candal_A/0/1/0/all/0/1">A. Alvarez-Candal</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Barbosa_C/0/1/0/all/0/1">C. L. Barbosa</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Castellon_J/0/1/0/all/0/1">J. L. N. Castell&#xf3;n</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Coelho_P/0/1/0/all/0/1">P. Coelho</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Dantas_M/0/1/0/all/0/1">M. L. L. Dantas</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Dupke_R/0/1/0/all/0/1">R. Dupke</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Ederoclite_A/0/1/0/all/0/1">A. Ederoclite</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Galarza_A/0/1/0/all/0/1">A. Galarza</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Goncalves_T/0/1/0/all/0/1">T. S. Gon&#xe7;alves</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Hernandez_Jimenez_J/0/1/0/all/0/1">J. A. Hernandez-Jimenez</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Jimenez_Teja_Y/0/1/0/all/0/1">Y. Jim&#xe9;nez-Teja</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Lopes_A/0/1/0/all/0/1">A. Lopes</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Lopes_P/0/1/0/all/0/1">P. A. A. Lopes</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Oliveira_R/0/1/0/all/0/1">R. Lopes de Oliveira</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Azevedo_J/0/1/0/all/0/1">J. L. Melo de Azevedo</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Nakazono_L/0/1/0/all/0/1">L. M. Nakazono</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Perottoni_H/0/1/0/all/0/1">H. D. Perottoni</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Queiroz_C/0/1/0/all/0/1">C. Queiroz</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Saha_K/0/1/0/all/0/1">K. Saha</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Sodre_L/0/1/0/all/0/1">L. Sodr&#xe9; Jr.</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Telles_E/0/1/0/all/0/1">E. Telles</a>, <a href="http://arxiv.org/find/astro-ph/1/au:+Souza_R/0/1/0/all/0/1">R. C. Thom de Souza</a>

We present a star/galaxy classification for the Southern Photometric Local
Universe Survey (S-PLUS), based on a Machine Learning approach: the Random
Forest algorithm. We train the algorithm using the S-PLUS optical photometry up
to $r$=21, matched to SDSS/DR13, and morphological parameters. The metric of
importance is defined as the relative decrease of the initial accuracy when all
correlations related to a certain feature is vanished. In general, the broad
photometric bands presented higher importance when compared to narrow ones. The
influence of the morphological parameters has been evaluated training the RF
with and without the inclusion of morphological parameters, presenting accuracy
values of 95.0% and 88.1%, respectively. Particularly, the morphological
parameter {rm FWHM/PSF} performed the highest importance over all features to
distinguish between stars and galaxies, indicating that it is crucial to
classify objects into stars and galaxies. We investigate the misclassification
of stars and galaxies in the broad-band colour-colour diagram $(g-r)$ versus
$(r-i)$. The morphology can notably improve the classification of objects at
regions in the diagram where the misclassification was relatively high.
Consequently, it provides cleaner samples for statistical studies. The expected
contamination rate of red galaxies as a function of the redshift is estimated,
providing corrections for red galaxy samples. The classification of QSOs as
extragalactic objects is slightly better using photometric-only case. An
extragalactic point-source catalogue is provided using the classification
without any morphology feature (only the SED information) with additional
constraints on photometric redshifts and {rm FWHM/PSF} values.

We present a star/galaxy classification for the Southern Photometric Local
Universe Survey (S-PLUS), based on a Machine Learning approach: the Random
Forest algorithm. We train the algorithm using the S-PLUS optical photometry up
to $r$=21, matched to SDSS/DR13, and morphological parameters. The metric of
importance is defined as the relative decrease of the initial accuracy when all
correlations related to a certain feature is vanished. In general, the broad
photometric bands presented higher importance when compared to narrow ones. The
influence of the morphological parameters has been evaluated training the RF
with and without the inclusion of morphological parameters, presenting accuracy
values of 95.0% and 88.1%, respectively. Particularly, the morphological
parameter {rm FWHM/PSF} performed the highest importance over all features to
distinguish between stars and galaxies, indicating that it is crucial to
classify objects into stars and galaxies. We investigate the misclassification
of stars and galaxies in the broad-band colour-colour diagram $(g-r)$ versus
$(r-i)$. The morphology can notably improve the classification of objects at
regions in the diagram where the misclassification was relatively high.
Consequently, it provides cleaner samples for statistical studies. The expected
contamination rate of red galaxies as a function of the redshift is estimated,
providing corrections for red galaxy samples. The classification of QSOs as
extragalactic objects is slightly better using photometric-only case. An
extragalactic point-source catalogue is provided using the classification
without any morphology feature (only the SED information) with additional
constraints on photometric redshifts and {rm FWHM/PSF} values.

http://arxiv.org/icons/sfx.gif