From 0fb02512addddd2774a0e3b2cae2cb74ed5bceee Mon Sep 17 00:00:00 2001 From: MauricioGarciaS <47052044+MauricioGarciaS@users.noreply.github.com> Date: Mon, 11 Dec 2023 13:38:29 +0000 Subject: [PATCH] chore(recommendations): Updated python modules for trainer. Added script to save features from feedback sessions into ml database. --- ee/recommendation/ml_trainer/airflow/dags/db_update.py | 5 +++-- ee/recommendation/ml_trainer/requirements.txt | 6 +++--- ee/recommendation/utils/ch_client.py | 2 ++ 3 files changed, 8 insertions(+), 5 deletions(-) diff --git a/ee/recommendation/ml_trainer/airflow/dags/db_update.py b/ee/recommendation/ml_trainer/airflow/dags/db_update.py index 53f1b868b..56385746e 100644 --- a/ee/recommendation/ml_trainer/airflow/dags/db_update.py +++ b/ee/recommendation/ml_trainer/airflow/dags/db_update.py @@ -30,9 +30,10 @@ dbname = config('pg_dbname_ml') password = config('pg_password_ml') tracking_uri = f"postgresql+psycopg2://{user}:{password}@{host}:{port}/{dbname}" - +# 1702296756 def get_today_feedback(): - query = "SELECT project_id, session_id, user_id as viewer_id, payload FROM recommendation_feedback" + current_datetime = int((datetime.now()-timedelta(seconds=execute_interval)).timestamp()) + query = f"SELECT project_id, session_id, user_id as viewer_id, payload FROM recommendation_feedback WHERE insertion_time >= {current_datetime}" connection_handler = ConnectionHandler(tracking_uri) # Connection to mlflow's database diff --git a/ee/recommendation/ml_trainer/requirements.txt b/ee/recommendation/ml_trainer/requirements.txt index 0a2b6cd6a..c48512bf6 100644 --- a/ee/recommendation/ml_trainer/requirements.txt +++ b/ee/recommendation/ml_trainer/requirements.txt @@ -1,3 +1,3 @@ -argcomplete==3.1.2 -apache-airflow==2.7.2 -airflow-code-editor==7.4.1 +argcomplete==3.2.1 +apache-airflow==2.7.3 +airflow-code-editor==7.5.0 diff --git a/ee/recommendation/utils/ch_client.py b/ee/recommendation/utils/ch_client.py index 514820212..1633dbed2 100644 --- a/ee/recommendation/utils/ch_client.py +++ b/ee/recommendation/utils/ch_client.py @@ -22,6 +22,8 @@ class ClickHouseClient: self.__client = clickhouse_driver.Client(host=config("ch_host"), database="default", port=config("ch_port", cast=int), + user=config("ch_user", cast=str), + password=config("ch_password", cast=str), settings=settings) \ if self.__client is None else self.__client