Created by Rafael S. de Souza on 26 May 2020

This folder contains pre-processed data from the
Supernova Photometric Classification Challenge (SNPCC):
https://arxiv.org/abs/1008.1024.

All data were feature extracted using the Bazin function:
https://arxiv.org/pdf/0904.1066.pdf

This data was used to obtain the results reported in 
Kennamer et al., 2020 - https://arxiv.org/pdf/2010.05941.pdf

The data has been separated into Training, Test, Validation and Pool.

***********************
*** realistic *********

In this experiment desing, Training follows the distribution of a
realistic spectroscopically observed data set and is highly biased
towards brighter and higher SNR objects, when compare to the Test
and Validation samples.
***********************

Training, Test and Validation samples
have 1000 objects each, and correspond to full light curve observations.

The pool sample was pre-processed separated for each
observation day and stored in a subfolder.

pool/
in this folder, each file, day_X.dat, corresponds to a given
day of observation X since the start of the survey.

the pool sample holds, beyond all features available to the
other samples, two extra columns: 'cost_4m' and 'cost_8m' this
indicates how many seconds are necessary to observe a particular
object in a 4m and 8m telescope, respectively.