In this notebook you will find instructions on how to run the star, quasar, galaxy classification for the S-PLUS survey.
Install the splusdata package (version >= 3.83) for Python 3+:
pip install splusdata
The SQGClass is a pre-trained estimator that works especifically for the S-PLUS survey. More information about the survey can be found in Mendes de Oliveira et al. 2019 and Almeida-Fernandes et al. 2022. For more information about the classification performances, please check Nakazono et al. 2021.
- model: {"auto", "RF18", "RF16"}, default = "auto"
Indicates whether to use a model that was pre-trained with the addition of infrared magnitudes (W1 and W2 from ALLWISE) or not.
If model = "auto", it automatically run the best model for each case. If model = "RF18", it only classify objects that have WISE counterpart. If model = "RF16", it classified all objects with a model that do not include infrared magnitudes in the per-trained model.
Please ensure that you are aware of the different expected performances for RF18 and RF16. For that, check Section 6 from Nakazono et al. 2021.
- verbose: True or False
Controls the verbosity when loading the pre-trained model
SQGClass has a method classify() to predict class and probabilities for a single object or a dataframe.
- CLASS: float
The class atributted to an object. If CLASS = 0, the object is classified as a quasar. If CLASS = 1, the object is classified as a star. If CLASS = 2, the object is classified as a galaxy.
- model_flag: int
Flag indicating whether the classification was obtained when training the model with the addition of infrared magnitudes (W1 and W2 from ALLWISE) or not. This is only returned when model = "auto". If model_flag = 0, ALLWISE magnitudes are included in the pre-trained model. The flag is 1, otherwise.
- PROB_QSO: float
Estimated probability of the object being a quasar [0,1]. Only returned when return_prob = True.
- PROB_STAR: float
Estimated probability of the object being a star [0,1]. Only returned when return_prob = True.
- PROB_GAL: float
Estimated probability of the object being a galaxy [0,1]. Only returned when return_prob = True.
- df: pandas dataframe
This pandas dataframe must contain the following features:
['u_iso', 'J0378_iso','J0395_iso', 'J0410_iso','J0430_iso','g_iso','J0515_iso', 'r_iso','J0660_iso', 'i_iso', 'J0861_iso','z_iso']If match_irsa = True, it also must contain
['RA', 'DEC']
- return_prob: True or False, default = True
Indicates if probabilities should be included in the output.
- match_irsa: True or False, default = False
If True, queries ALLWISE observations from IRSA TAP service. Attention: if this is set to True, ensure that you enter a limited set of objects each time you call this method, otherwise it becomes computationally expensive to perform the query and the cross-match. We recommend that you classify objects from a single S-PLUS field per time.
- columns_wise: dict, default = {'w1mpro': 'w1mpro','w2mpro': 'w2mpro', 'w1snr': 'w1snr','w2snr': 'w2snr','w1sigmpro': 'w1sigmpro', 'w2sigmpro': 'w2sigmpro'}
Sets the names of each ALLWISE column in your df. For instance, if W1 is named "w1_mag" in your dataframe, do as follows: {'w1mpro': 'w1_mag',...}
- verbose = True or False, default = False
Controls verbosity
If you have any problem, please feel free to contact me or open an issue.