I've noticed that some of the top Pub LB notebooks have either a single row or empty submissions.
\nAm I to assume that this is due to this being a code competition, and test sets are switched out?
\nSome notebooks do have a fleshed out submission file, so slightly confused here.
I've noticed that some of the top Pub LB notebooks have either a single row or empty submissions.
\nAm I to assume that this is due to this being a code competition, and test sets are switched out?
\nSome notebooks do have a fleshed out submission file, so slightly confused here.
since this is a code competition, the notebooks output that you see after comitting, is not on the original test data, but just a sample of how test data looks like, containing very few steps and very few series ids. In code competitions you will not have access to the actual test data. The sample test data is replaced with original test data by kaggle, once you submit the notebook. Since the models don't detect any wakeup or sleep activity in the period for the sample test data, hence few or no rows.
", "rawMarkdown": "since this is a code competition, the notebooks output that you see after comitting, is not on the original test data, but just a sample of how test data looks like, containing very few steps and very few series ids. In code competitions you will not have access to the actual test data. The sample test data is replaced with original test data by kaggle, once you submit the notebook. Since the models don't detect any wakeup or sleep activity in the period for the sample test data, hence few or no rows.", "votes": null }, { "id": "2546063", "postDate": "12/02/2023 04:47:48", "content": "Thank you, I had a hunch it was something like this. However, some pub notebooks still produce data filled submissions, hence my confusion.
\nhttps://www.kaggle.com/code/danielphalen/cmss-grunet-infer/output?select=submission.csv
\nI would guess it has something to do with the following check:
\n`test = pd.read_parquet(PATHS.TEST_SERIES)
\nseries_ids = list(test.series_id.unique())
\nif len(series_ids) == 3:
\n print('Is test script')
\n test = pd.read_parquet(PATHS.TRAIN_SERIES)
\n series_ids = list(test.series_id.unique())[:10]
len(series_ids)`
", "rawMarkdown": "Thank you, I had a hunch it was something like this. However, some pub notebooks still produce data filled submissions, hence my confusion.\n\nhttps://www.kaggle.com/code/danielphalen/cmss-grunet-infer/output?select=submission.csv\n\nI would guess it has something to do with the following check:\n\n`test = pd.read_parquet(PATHS.TEST_SERIES)\nseries_ids = list(test.series_id.unique())\nif len(series_ids) == 3:\n print('Is test script')\n test = pd.read_parquet(PATHS.TRAIN_SERIES)\n series_ids = list(test.series_id.unique())[:10]\n \nlen(series_ids)`", "votes": null } ], "comments": [ { "id": 2545605, "author_name": "nikhilmishradev", "author_url": "", "post_date": "12/01/2023 15:48:58", "content": "since this is a code competition, the notebooks output that you see after comitting, is not on the original test data, but just a sample of how test data looks like, containing very few steps and very few series ids. In code competitions you will not have access to the actual test data. The sample test data is replaced with original test data by kaggle, once you submit the notebook. Since the models don't detect any wakeup or sleep activity in the period for the sample test data, hence few or no rows.
", "votes": null, "replies": [ { "id": 2546063, "author_name": "exjustice", "author_url": "", "post_date": "12/02/2023 04:47:48", "content": "Thank you, I had a hunch it was something like this. However, some pub notebooks still produce data filled submissions, hence my confusion.
\nhttps://www.kaggle.com/code/danielphalen/cmss-grunet-infer/output?select=submission.csv
\nI would guess it has something to do with the following check:
\n`test = pd.read_parquet(PATHS.TEST_SERIES)
\nseries_ids = list(test.series_id.unique())
\nif len(series_ids) == 3:
\n print('Is test script')
\n test = pd.read_parquet(PATHS.TRAIN_SERIES)
\n series_ids = list(test.series_id.unique())[:10]
len(series_ids)`
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2545269": "I've noticed that some of the top Pub LB notebooks have either a single row or empty submissions.\nAm I to assume that this is due to this being a code competition, and test sets are switched out?\nSome notebooks do have a fleshed out submission file, so slightly confused here.", "2545605": "since this is a code competition, the notebooks output that you see after comitting, is not on the original test data, but just a sample of how test data looks like, containing very few steps and very few series ids. In code competitions you will not have access to the actual test data. The sample test data is replaced with original test data by kaggle, once you submit the notebook. Since the models don't detect any wakeup or sleep activity in the period for the sample test data, hence few or no rows.", "2546063": "Thank you, I had a hunch it was something like this. However, some pub notebooks still produce data filled submissions, hence my confusion.\n\nhttps://www.kaggle.com/code/danielphalen/cmss-grunet-infer/output?select=submission.csv\n\nI would guess it has something to do with the following check:\n\n`test = pd.read_parquet(PATHS.TEST_SERIES)\nseries_ids = list(test.series_id.unique())\nif len(series_ids) == 3:\n print('Is test script')\n test = pd.read_parquet(PATHS.TRAIN_SERIES)\n series_ids = list(test.series_id.unique())[:10]\n \nlen(series_ids)`" }, "source": "meta" }