![\"start_carts_without_click\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2743215%2F992e0ea92ca655b919a6603be19f1765%2Fstart_carts.png?generation=1669288124642750&alt=media\")
The train dataset starts from 2022-07-31 22:00:00 from the
\n\"otto-full-optimized-memory-footprint\" and \"otto-chunk-data-inparquet-format\" datasets
I was trying to create a experiment dataset for sequence-prediction, just like any RNN or Transformers architecture.
\nThose get insight from \"previous behaviors\". However, there are many sessions that starts with \"carts\".
The type mapping is followed, {'clicks': 0, 'carts': 1, 'orders': 2}
\n
My hypothesis was the session \"clicks\" are related what the user in the session tried to search a item (aid), so the clicks might be affect to the user's \"carts or orders\", as cause and effect problem.
\nTo resolve and create a proper sequences, may need some preprocessing steps for filling, padding or dropping.
\nchecking notebook: here
", "messages": [ { "id": 2042010, "postDate": "2022-11-24T11:28:20.380Z", "content": "The train dataset starts from 2022-07-31 22:00:00 from the
\n\"otto-full-optimized-memory-footprint\" and \"otto-chunk-data-inparquet-format\" datasets
I was trying to create a experiment dataset for sequence-prediction, just like any RNN or Transformers architecture.
\nThose get insight from \"previous behaviors\". However, there are many sessions that starts with \"carts\".
The type mapping is followed, {'clicks': 0, 'carts': 1, 'orders': 2}
\n
My hypothesis was the session \"clicks\" are related what the user in the session tried to search a item (aid), so the clicks might be affect to the user's \"carts or orders\", as cause and effect problem.
\nTo resolve and create a proper sequences, may need some preprocessing steps for filling, padding or dropping.
\nchecking notebook: here
", "rawMarkdown": "The train dataset starts from 2022-07-31 22:00:00 from the\n\"[otto-full-optimized-memory-footprint](https://www.kaggle.com/datasets/radek1/otto-full-optimized-memory-footprint)\" and \"[otto-chunk-data-inparquet-format](https://www.kaggle.com/datasets/columbia2131/otto-chunk-data-inparquet-format)\" datasets\n\nI was trying to create a experiment dataset for sequence-prediction, just like any RNN or Transformers architecture.\nThose get insight from \"previous behaviors\". However, there are many sessions that starts with \"carts\".\n\nThe type mapping is followed, {'clicks': 0, 'carts': 1, 'orders': 2}\n\n\n\n\nMy hypothesis was the session \"clicks\" are related what the user in the session tried to search a item (aid), so the clicks might be affect to the user's \"carts or orders\", as cause and effect problem. \n\nTo resolve and create a proper sequences, may need some preprocessing steps for filling, padding or dropping. \n\nchecking notebook: [here](https://www.kaggle.com/seunghwan1228/cheking-oov-test-aid)", "votes": 2 } ], "comments": [], "raw_markdown_by_id": { "2042010": "The train dataset starts from 2022-07-31 22:00:00 from the\n\"[otto-full-optimized-memory-footprint](https://www.kaggle.com/datasets/radek1/otto-full-optimized-memory-footprint)\" and \"[otto-chunk-data-inparquet-format](https://www.kaggle.com/datasets/columbia2131/otto-chunk-data-inparquet-format)\" datasets\n\nI was trying to create a experiment dataset for sequence-prediction, just like any RNN or Transformers architecture.\nThose get insight from \"previous behaviors\". However, there are many sessions that starts with \"carts\".\n\nThe type mapping is followed, {'clicks': 0, 'carts': 1, 'orders': 2}\n\n\n\n\nMy hypothesis was the session \"clicks\" are related what the user in the session tried to search a item (aid), so the clicks might be affect to the user's \"carts or orders\", as cause and effect problem. \n\nTo resolve and create a proper sequences, may need some preprocessing steps for filling, padding or dropping. \n\nchecking notebook: [here](https://www.kaggle.com/seunghwan1228/cheking-oov-test-aid)" } }