Merge pull request #155 from george0st/change

Decrease amount of transaction and events
george0st · Apr 18, 2024 · 67a258b · 67a258b
2 parents 6398032 + c0a15b5
commit 67a258b
Show file tree

Hide file tree

Showing 32 changed files with 595 additions and 557 deletions.
diff --git a/02-data/01-size-100/01-basic-party.csv.gz b/02-data/01-size-100/01-basic-party.csv.gz
diff --git a/02-data/01-size-100/01-basic-party.parquet b/02-data/01-size-100/01-basic-party.parquet
diff --git a/02-data/01-size-100/02-basic-contact.csv.gz b/02-data/01-size-100/02-basic-contact.csv.gz
diff --git a/02-data/01-size-100/02-basic-contact.parquet b/02-data/01-size-100/02-basic-contact.parquet
diff --git a/02-data/01-size-100/03-basic-relation.csv.gz b/02-data/01-size-100/03-basic-relation.csv.gz
diff --git a/02-data/01-size-100/03-basic-relation.parquet b/02-data/01-size-100/03-basic-relation.parquet
diff --git a/02-data/01-size-100/04-basic-account.csv.gz b/02-data/01-size-100/04-basic-account.csv.gz
diff --git a/02-data/01-size-100/04-basic-account.parquet b/02-data/01-size-100/04-basic-account.parquet
diff --git a/02-data/01-size-100/05-basic-transaction.csv.gz b/02-data/01-size-100/05-basic-transaction.csv.gz
diff --git a/02-data/01-size-100/05-basic-transaction.parquet b/02-data/01-size-100/05-basic-transaction.parquet
diff --git a/02-data/01-size-100/06-basic-event.csv.gz b/02-data/01-size-100/06-basic-event.csv.gz
diff --git a/02-data/01-size-100/06-basic-event.parquet b/02-data/01-size-100/06-basic-event.parquet
diff --git a/02-data/01-size-100/07-basic-communication.csv.gz b/02-data/01-size-100/07-basic-communication.csv.gz
diff --git a/02-data/01-size-100/07-basic-communication.parquet b/02-data/01-size-100/07-basic-communication.parquet
diff --git a/02-data/02-size-1K/01-basic-party.csv.gz b/02-data/02-size-1K/01-basic-party.csv.gz
diff --git a/02-data/02-size-1K/01-basic-party.parquet b/02-data/02-size-1K/01-basic-party.parquet
diff --git a/02-data/02-size-1K/02-basic-contact.csv.gz b/02-data/02-size-1K/02-basic-contact.csv.gz
diff --git a/02-data/02-size-1K/02-basic-contact.parquet b/02-data/02-size-1K/02-basic-contact.parquet
diff --git a/02-data/02-size-1K/03-basic-relation.csv.gz b/02-data/02-size-1K/03-basic-relation.csv.gz
diff --git a/02-data/02-size-1K/03-basic-relation.parquet b/02-data/02-size-1K/03-basic-relation.parquet
diff --git a/02-data/02-size-1K/04-basic-account.csv.gz b/02-data/02-size-1K/04-basic-account.csv.gz
diff --git a/02-data/02-size-1K/04-basic-account.parquet b/02-data/02-size-1K/04-basic-account.parquet
diff --git a/02-data/02-size-1K/05-basic-transaction.csv.gz b/02-data/02-size-1K/05-basic-transaction.csv.gz
diff --git a/02-data/02-size-1K/05-basic-transaction.parquet b/02-data/02-size-1K/05-basic-transaction.parquet
diff --git a/02-data/02-size-1K/06-basic-event.csv.gz b/02-data/02-size-1K/06-basic-event.csv.gz
diff --git a/02-data/02-size-1K/06-basic-event.parquet b/02-data/02-size-1K/06-basic-event.parquet
diff --git a/02-data/02-size-1K/07-basic-communication.csv.gz b/02-data/02-size-1K/07-basic-communication.csv.gz
diff --git a/02-data/02-size-1K/07-basic-communication.parquet b/02-data/02-size-1K/07-basic-communication.parquet
diff --git a/03-test/01-size-100.json b/03-test/01-size-100.json
diff --git a/03-test/02-size-1k.json b/03-test/02-size-1k.json
diff --git a/generator/basic_event.py b/generator/basic_event.py
@@ -80,18 +80,18 @@ def generate(self, count):
                 #   for customer:       more active
                 #   for non customer:   small amount of activities
                 if party_customer:
-                    day = self.rnd_choose(range(10),[0.01, 0.19, 0.1, 0.2, 0.1, 0.05, 0.05, 0.1, 0.1, 0.1])
+                    day = self.rnd_choose(range(10),[0, 0, 0, 0.1, 0.3, 0.15, 0.15, 0.1, 0.1, 0.1])
                 else:
                     day = self.rnd_choose(range(10), [0, 0, 0, 0, 0.05, 0.05, 0.1, 0.2, 0.3, 0.3])
                 event_date = event_date + datetime.timedelta(days=float(day))
                 if event_date > self.now:
                     break
 
                 # define bundle
-                #   for customer:       size 2-15x events (bigger amount of activities)
-                #   for non-customer:   size 2-10x events (small amount of activites)
+                #   for customer:       size 2-10x events (bigger amount of activities)
+                #   for non-customer:   size 2-5x events (small amount of activites)
                 session_id = str(uuid.uuid4())
-                session_events=self.rnd_choose(range(2, 15)) if party_customer else self.rnd_choose(range(2, 10))
+                session_events=self.rnd_choose(range(2, 10)) if party_customer else self.rnd_choose(range(2, 5))
                 session_datetime = datetime.datetime(event_date.year,
                                                      event_date.month,
                                                      event_date.day,

diff --git a/generator/basic_transaction.py b/generator/basic_transaction.py
@@ -10,6 +10,7 @@
 class BasicTransaction(BaseData):
 
     NAME= "05-basic-transaction"
+    MAX_EVENT_HISTORY_MONTHS = 3*12     # 3 years as default, in case of value '0' or '-1', it is without limit
 
     def __init__(self, path, gmodel):
         super().__init__(path, gmodel, BasicTransaction.NAME)
@@ -41,10 +42,15 @@ def generate(self, count):
 
             dif_date=round((date_to-date_from).days/30)
 
+            # apply limit for max history
+            if self.MAX_EVENT_HISTORY_MONTHS > 0:
+                if dif_date > self.MAX_EVENT_HISTORY_MONTHS:
+                    dif_date = self.MAX_EVENT_HISTORY_MONTHS
+
             for mounth in range(dif_date):
                 a=mounth*30
                 b=int(self.rnd_int(1,30))
-                new_date=date_from+datetime.timedelta(days=a+b)
+                new_date=date_to-datetime.timedelta(days=a+b)
 
                 model=self.model_item()