|
@@ -57,7 +57,7 @@ class DataProcessor(object):
|
|
|
label_dt = generate_label_date(hourly_dt_str)
|
|
|
label_obj = label_info.get(label_dt, {}).get(video_id)
|
|
|
if label_obj:
|
|
|
- label = int(label_obj["total_return"]) if label_obj["total_return"] else 0
|
|
|
+ label = int(label_obj["uplevel"]) if label_obj["uplevel"] else 0
|
|
|
# print(label)
|
|
|
else:
|
|
|
label = 0
|
|
@@ -71,14 +71,14 @@ class DataProcessor(object):
|
|
|
video_id, hour_dt = item_
|
|
|
# print(type(video_id))
|
|
|
label_info = self.label_data
|
|
|
- title = read_title(client=self.client, video_id=video_id)
|
|
|
+ # title = read_title(client=self.client, video_id=video_id)
|
|
|
label, dt_daily = generate_label(str(video_id), hour_dt, label_info)
|
|
|
|
|
|
- insert_sql = f"""UPDATE lightgbm_data set video_title = '{title}', label = '{label}', daily_dt_str = '{dt_daily}' where video_id = '{video_id}';"""
|
|
|
+ insert_sql = f"""UPDATE lightgbm_data set label = '{label}', daily_dt_str = '{dt_daily}' where video_id = '{video_id}';"""
|
|
|
# print(insert_sql)
|
|
|
self.client_spider.update(insert_sql)
|
|
|
|
|
|
- select_sql = "SELECT video_id, hour_dt_str FROM lightgbm_data where label is NULL;"
|
|
|
+ select_sql = "SELECT video_id, hour_dt_str FROM lightgbm_data;"
|
|
|
init_data_tuple = self.client_spider.select(select_sql)
|
|
|
init_list = list(init_data_tuple)
|
|
|
for item in tqdm(init_list):
|