pipe_config.yaml

data_path: "./youtube/GBvideos.csv"
category_map_path: "./youtube/GB_category_id.json"

# Pre-processing config
date_column_format:
  trending_date: "%y.%d.%m"
  publish_time: "%Y-%m-%dT%H:%M:%S%.fZ"

# Feature engineering config
ratio_features:
  likes_to_dislikes:
    - likes
    - dislikes
  likes_to_views:
    - likes
    - views
  comments_to_views:
    - comment_count
    - views

difference_features:
  days_to_trending:
    - trending_date
    - publish_time

date_features:
  trending_date:
    - weekday

# Filtering config
max_time_to_trending: 60

# Features config
join_columns:
  - video_id
  - trending_date

base_columns:
  - views
  - likes
  - dislikes
  - comment_count
  - comments_disabled
  - ratings_disabled
  - video_error_or_removed
  - likes_to_dislikes
  - likes_to_views
  - comments_to_views
  - trending_date_weekday
  - channel_title
  - tags
  - description
  - category_id

aggregate_windows:
  - 7
  - 30
  - 180