{ "id": 366169, "title": "How events in test sessions are split?", "url": "/competitions/otto-recommender-system/discussion/366169", "author_name": "", "post_date": "2022-11-15T06:17:06.323102800Z", "votes": 5, "comment_count": 3, "views": 0, "content": "

@pnormann

\n

What is the split method of test sessions to calculate public & private LBs?

\n

Possibilities:

\n\n

\"Screen-Shot-2022-11-15-at-15-09-45\"

", "messages": [ { "id": "2030017", "postDate": "11/15/2022 06:17:06", "content": "

@pnormann

\n

What is the split method of test sessions to calculate public & private LBs?

\n

Possibilities:

\n\n

\"Screen-Shot-2022-11-15-at-15-09-45\"

", "rawMarkdown": "pnormann\n\nWhat is the split method of test sessions to calculate public & private LBs?\n\nPossibilities:\n* number of events (if a session has N events in total, provided evens are first (1-p) * N events, and hidden evens are last p * N events) \n* term: (if a session has n days long, provided events are events in first (1-p) * n days, and hidden evens are the events in the last p * N days)\n* etc.\n\n\"Screen-Shot-2022-11-15-at-15-09-45\"", "votes": null }, { "id": "2030126", "postDate": "11/15/2022 08:01:59", "content": "

We can let Philipp reply, but i think the code to split public private is here. It looks like a random split of events

\n
test_events = ground_truth(deepcopy(events))\nif not split_idx:\n    split_idx = random.randint(1, len(test_events))\ntest_events = test_events[:split_idx]\n
", "rawMarkdown": "We can let Philipp reply, but i think the code to split public private is [here][1]. It looks like a random split of events\n\n test_events = ground_truth(deepcopy(events))\n if not split_idx:\n split_idx = random.randint(1, len(test_events))\n test_events = test_events[:split_idx]\n\n[1]: https://github.com/otto-de/recsys-dataset/blob/f5bf55d4bf9c1e9be367563289a2beb04a9b9b58/src/testset.py#L22", "votes": null }, { "id": "2030127", "postDate": "11/15/2022 08:02:35", "content": "

Hi,
\nyou should refer to the github repo of the competition to answer this question.

\n

In particular the link above in this comment takes you to the split_events function that shows you the answer. The split point is decided over the session lenght choosing a random split point with random.randint(1, len(test_events)) , where test_event contains a single session as a list of user actions.

", "rawMarkdown": "Hi,\nyou should refer to the github [repo](https://github.com/otto-de/recsys-dataset/blob/f5bf55d4bf9c1e9be367563289a2beb04a9b9b58/src/testset.py#L22) of the competition to answer this question.\n\nIn particular the link above in this comment takes you to the **split_events** function that shows you the answer. The split point is decided over the session lenght choosing a random split point with **random.randint(1, len(test_events))** , where test_event contains a single session as a list of user actions.", "votes": null }, { "id": "2030156", "postDate": "11/15/2022 08:29:53", "content": "

@pietromaldini1 @cdeotte Thanks for the reply.

", "rawMarkdown": "pietromaldini1 @cdeotte Thanks for the reply.", "votes": null } ], "comments": [ { "id": 2030126, "author_name": "cdeotte", "author_url": "", "post_date": "11/15/2022 08:01:59", "content": "

We can let Philipp reply, but i think the code to split public private is here. It looks like a random split of events

\n
test_events = ground_truth(deepcopy(events))\nif not split_idx:\n    split_idx = random.randint(1, len(test_events))\ntest_events = test_events[:split_idx]\n
", "votes": null, "replies": [] }, { "id": 2030127, "author_name": "pietromaldini1", "author_url": "", "post_date": "11/15/2022 08:02:35", "content": "

Hi,
\nyou should refer to the github repo of the competition to answer this question.

\n

In particular the link above in this comment takes you to the split_events function that shows you the answer. The split point is decided over the session lenght choosing a random split point with random.randint(1, len(test_events)) , where test_event contains a single session as a list of user actions.

", "votes": null, "replies": [] }, { "id": 2030156, "author_name": "tatamikenn", "author_url": "", "post_date": "11/15/2022 08:29:53", "content": "

@pietromaldini1 @cdeotte Thanks for the reply.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2030017": "pnormann\n\nWhat is the split method of test sessions to calculate public & private LBs?\n\nPossibilities:\n* number of events (if a session has N events in total, provided evens are first (1-p) * N events, and hidden evens are last p * N events) \n* term: (if a session has n days long, provided events are events in first (1-p) * n days, and hidden evens are the events in the last p * N days)\n* etc.\n\n\"Screen-Shot-2022-11-15-at-15-09-45\"", "2030126": "We can let Philipp reply, but i think the code to split public private is [here][1]. It looks like a random split of events\n\n test_events = ground_truth(deepcopy(events))\n if not split_idx:\n split_idx = random.randint(1, len(test_events))\n test_events = test_events[:split_idx]\n\n[1]: https://github.com/otto-de/recsys-dataset/blob/f5bf55d4bf9c1e9be367563289a2beb04a9b9b58/src/testset.py#L22", "2030127": "Hi,\nyou should refer to the github [repo](https://github.com/otto-de/recsys-dataset/blob/f5bf55d4bf9c1e9be367563289a2beb04a9b9b58/src/testset.py#L22) of the competition to answer this question.\n\nIn particular the link above in this comment takes you to the **split_events** function that shows you the answer. The split point is decided over the session lenght choosing a random split point with **random.randint(1, len(test_events))** , where test_event contains a single session as a list of user actions.", "2030156": "pietromaldini1 @cdeotte Thanks for the reply." }, "source": "meta" }