minor...

modules/icing/pirep_goes.py

@@ -366,6 +366,7 @@ def analyze(ice_dct, no_ice_dct):
         print(dt_str[2:])
 
 
+# This mostly reduces some categories for a degree of class balancing
 def process(ice_dct, no_ice_dct, neg_ice_dct):
     new_ice_dct = {}
     new_no_ice_dct = {}
@@ -446,14 +447,16 @@ def process(ice_dct, no_ice_dct, neg_ice_dct):
     ice_tidx_2 = ice_tidx_2[ridxs]
     ice_keys_2 = ice_keys_2[::2]
     ice_tidx_2 = ice_tidx_2[::2]
+    print('2: reduced: ', ice_tidx_2.shape)
 
     ice_keys_1 = np.array(ice_keys_1)
     ice_tidx_1 = np.array(ice_tidx_1)
     print('1: ', ice_keys_1.shape[0])
-    print('no intensity: ', no_intensity_cnt)
+    print('0: ', no_intensity_cnt)
 
     ice_keys = np.concatenate([ice_keys_5_6, ice_keys_1, ice_keys_2, ice_keys_3, ice_keys_4])
     ice_tidx = np.concatenate([ice_tidx_5_6, ice_tidx_1, ice_tidx_2, ice_tidx_3, ice_tidx_4])
+    print('icing total reduced: ', ice_tidx.shape)
 
     sidxs = np.argsort(ice_keys)
     ice_keys = ice_keys[sidxs]
@@ -479,7 +482,7 @@ def process(ice_dct, no_ice_dct, neg_ice_dct):
     no_ice_tidx = no_ice_tidx[ridxs]
     no_ice_keys = no_ice_keys[::20]
     no_ice_tidx = no_ice_tidx[::20]
-    print('no ice   : ', no_ice_keys.shape[0])
+    print('no ice reduced: ', no_ice_keys.shape[0])
 
     sidxs = np.argsort(no_ice_keys)
     no_ice_keys = no_ice_keys[sidxs]
@@ -500,9 +503,9 @@ def process(ice_dct, no_ice_dct, neg_ice_dct):
     print('neg ice total: ', neg_ice_keys.shape[0])
     np.random.seed(42)
     np.random.shuffle(neg_ice_keys)
-    neg_ice_keys = neg_ice_keys[0:10000]
+    neg_ice_keys = neg_ice_keys[0:12000]
     uniq_sorted_neg_ice = np.unique(neg_ice_keys)
-    print('neg ice: ', neg_ice_keys.shape, uniq_sorted_neg_ice.shape)
+    print('neg ice reduced: ', uniq_sorted_neg_ice.shape)
 
     for key in uniq_sorted_neg_ice:
         new_neg_ice_dct[key] = neg_ice_dct[key]