# Dataset information: the entry point for dataset configuration, including file locations for each modality.
dataset_id: M5
dataset_name: M5 Forecasting Accuracy
domain: retail_demand_forecasting

sample_defination:
  sample_type: item-store sales episode
  sample_id_graph: episode_id
  description: each sample corresponds to one item sold at one Walmart store

modalities:
# Available feature modalities.
  meta_features:
    required: true
    file_name: meta.csv

  node_level_task_graphs:
    available: true
    graph_ids:
      - item_graph

  time_series:
    available: true
    time_series_ids:
      - calendar_month
      - calendar_wday
      - calendar_year
      - sell_price
      - snap_CA
      - snap_TX
      - snap_WI
      - unit_sales
    
tasks:
# Supported task types, for example meta_feature prediction, time series forecasting, and ranking.
  - item_category_classification
  - episode_unit_sales_forecasting

# Data source information.
license:
  data_license: Kaggle competition data terms
  redistribution_allowed: source-dependent

source:
  name: M5 Forecasting - Accuracy
  url: https://www.kaggle.com/competitions/m5-forecasting-accuracy