multi hot

tensnsorflow/feature_engineering.py

@@ -78,22 +78,6 @@ def get_data():
     return validate_date,value_map
 
 
-def multi_hot(df,i,n):
-    ItemID_set = set()
-    for i in df[i].unique():
-        ItemID_set.update(set(i.split(",")))
-    ItemID2int = dict(zip(list(ItemID_set),list(range(n+1,n+1+len(ItemID_set),1))))
-    ItemID_map = {val: [ItemID2int[row] for row in val.split(',')] \
-                  for ii, val in enumerate(set(df[1]))}
-    ItemID_map_max_len = 3
-    for key in ItemID_map:
-        for cnt in range(ItemID_map_max_len - len(ItemID_map[key])):
-            ItemID_map[key].insert(len(ItemID_map[key]) + cnt, 88)
-    df[i] = df[i].map(ItemID_map)
-
-
-
-
 def write_csv(df,name,n):
     for i in range(0, df.shape[0], n):
         if i == 0:

tensnsorflow/multi_hot.py

@@ -26,7 +26,7 @@ def multi():
     df["l2"] = "lost"
     df["l3"] = "lost"
     for i in list(df["level"].unique()):
-        l = i.split(",")
+        l = i.split(";")
         if len(l) == 3:
             df.loc[df["level"] == i, ["l1"]] = l[0]
             df.loc[df["level"] == i, ["l2"]] = l[1]