修改测试文件

tensnsorflow/multi.py

@@ -18,9 +18,7 @@ def app_list_func(x,l):
     return ",".join([str(j) for j in e])
 
 def multi_hot(df,column,n):
-    df = df.select(column).fillna("lost_na")
-    df.show(6)
-    app_list_value = [i.split(",") for i in df.select(column).unique()]
+    app_list_value = [i.split(",") for i in df.select(column).collect().unique()]
     app_list_unique = []
     for i in app_list_value:
         app_list_unique.extend(i)
@@ -80,8 +78,10 @@ def feature_engineer():
     df = df.drop_duplicates(["ucity_id", "level2_ids", "ccity_name", "device_type", "manufacturer",
                               "channel", "top", "time", "stat_date", "app_list", "hospital_id", "level3_ids"])
 
-
-    multi_hot(df, "app_list", 1)
+    df = df.fillna("na")
+    df = df.toPandas()
+    print(df.shape)
+    print(df.head(6))