Vetting the Training Set performance



In [1]:

    
# imports
from linetools.spectra import io as lsio

Load spec



In [2]:

    
spec96451 = lsio.readspec(os.getenv('DROPBOX_DIR')+'/MachineLearning/DLAs/training_96451_5000.hdf5')

Check false positives



In [20]:

    
#idx = 79 #-- JXP bug??
#idx = 68 #-- Lyb
#idx = 67  #-- Noisy JXP
#idx = 30 # Check JXP
#idx = 4903 # Not compelling
#idx = 4992 # JXP gave 20.25
#idx = 4927 # JXP gave 20.15
#idx = 4985 # JXP never triggered
idx = 4661 # JXP never triggered
idx = 4882 # JXP never triggered
idx = 4775 # JXP didn't trigger (looks ok)
idx = 4709 # Quite possible strong DLA (with metals)
idx = 4679



In [21]:

    
spec96451.meta['headers'][idx]['PLATE'], spec96451.meta['headers'][idx]['FIBER']









    Out[21]:





(840, 126)



In [7]:

    
4614.*(1025.722/1215.67)









    Out[7]:





3893.0641605040837



In [14]:

    
4.1104*1215.67









    Out[14]:





4996.889968



In [ ]: