Hi,
\nI am losing my mind trying to make a submission but to no avail. Is there anyway I can debug what is wrong with my notebook? I checked out the parquet file that gets generated from the test set and it looks totally fine. Is there anything else I can do?
Hi,
\nI am losing my mind trying to make a submission but to no avail. Is there anyway I can debug what is wrong with my notebook? I checked out the parquet file that gets generated from the test set and it looks totally fine. Is there anything else I can do?
This is it https://www.kaggle.com/code/moetasimrady/fork-of-cmi-2025-first-submission
", "rawMarkdown": "This is it https://www.kaggle.com/code/moetasimrady/fork-of-cmi-2025-first-submission", "votes": null }, { "id": "3222989", "postDate": "06/12/2025 18:16:50", "content": "Kaggle Notebook is down :( . I dont know if i am the one facing the problem or is it face by someone else
", "rawMarkdown": "Kaggle Notebook is down :( . I dont know if i am the one facing the problem or is it face by someone else", "votes": null }, { "id": "3223397", "postDate": "06/13/2025 08:41:19", "content": "@sohier Can you please confirm?
", "rawMarkdown": "sohier Can you please confirm?", "votes": null }, { "id": "3223857", "postDate": "06/14/2025 02:17:17", "content": "This could be a result of not handling missing data in the hidden test case. One sanity check could be to run your submission script on the training data.
", "rawMarkdown": "This could be a result of not handling missing data in the hidden test case. One sanity check could be to run your submission script on the training data.", "votes": null }, { "id": "3224655", "postDate": "06/15/2025 09:14:23", "content": "Is it possible for some columns to not be included in the test set?
", "rawMarkdown": "Is it possible for some columns to not be included in the test set?", "votes": null }, { "id": "3225659", "postDate": "06/16/2025 17:33:58", "content": "It is possible that some null values may exist in the test set; however, the columns will remain identical to those in the provided test.csv file.
", "rawMarkdown": "It is possible that some null values may exist in the test set; however, the columns will remain identical to those in the provided test.csv file.", "votes": null }, { "id": "3225660", "postDate": "06/16/2025 17:34:39", "content": "Can't see it now😂 Problem solved?
", "rawMarkdown": "Can't see it now😂 Problem solved?", "votes": null }, { "id": "3225710", "postDate": "06/16/2025 19:03:04", "content": "I noticed a noob mistake, so I made it private out of shame, but actually after I fixed the noob mistake, it still won't submit, would very much appreciate taking a look 😁
", "rawMarkdown": "I noticed a noob mistake, so I made it private out of shame, but actually after I fixed the noob mistake, it still won't submit, would very much appreciate taking a look 😁", "votes": null }, { "id": "3225828", "postDate": "06/16/2025 22:37:20", "content": "You may want to check the DataFrame alignment. If penultimate_df or cleaned_df do not have perfectly matching keys in grouping_col, the merge could drop rows or produce unintended duplicates, especially after using reset_index().
For vanilla_categorical_procs and vanilla_quant_procs, if the test data contains unseen categories or different numeric ranges compared to the training data, the transformations may fail or yield inconsistent results.
Additionally, please double-check that the final output type is a string. It may also be beneficial to load the models and encoders before calling the predict function for improved efficiency.
You may debug using the below code after your predict function:
\nsequence_id = 'SEQ_000079' # 'SEQ_000033'
test_df = pl.read_csv('/kaggle/input/cmi-detect-behavior-with-sensor-data/train.csv')
\ntest_demo_df = pl.read_csv('/kaggle/input/cmi-detect-behavior-with-sensor-data/train_demographics.csv')
test_sequence = test_df.filter(pl.col('sequence_id') == sequence_id)
subject = test_sequence['subject'].to_list()[0]
\ntest_demo = test_demo_df.filter(pl.col('subject') == subject)
prediction = predict(test_sequence, test_demo)
\nprint(f\"sequence id: {sequence_id} | subject: {subject}\\nPrediction: {prediction}\")
see Below
\nExpect approximately 3,500 sequences in the hidden test set.
\nHalf of the hidden-test sequences are recorded with IMU only; the thermopile (thm_) and time-of-flight (tof__v*) columns are still present but contain null values for those sequences.
\nThis will allow us to determine whether adding the time-of-flight and thermopile sensors improves our ability to detect BFRBs. Note also that there is known sensor communication failure in this dataset, resulting in missing data from some sensors in some sequences.
Solved! It apprears that I only needed a single (df.fillna(0)) statement to handle NaN values
", "rawMarkdown": "Solved! It apprears that I only needed a single (df.fillna(0)) statement to handle NaN values", "votes": null } ], "comments": [ { "id": 3222734, "author_name": "moetasimrady", "author_url": "", "post_date": "06/12/2025 12:59:44", "content": "This is it https://www.kaggle.com/code/moetasimrady/fork-of-cmi-2025-first-submission
", "votes": null, "replies": [ { "id": 3225660, "author_name": "hearttt", "author_url": "", "post_date": "06/16/2025 17:34:39", "content": "Can't see it now😂 Problem solved?
", "votes": null, "replies": [ { "id": 3225710, "author_name": "moetasimrady", "author_url": "", "post_date": "06/16/2025 19:03:04", "content": "I noticed a noob mistake, so I made it private out of shame, but actually after I fixed the noob mistake, it still won't submit, would very much appreciate taking a look 😁
", "votes": null, "replies": [ { "id": 3225828, "author_name": "hearttt", "author_url": "", "post_date": "06/16/2025 22:37:20", "content": "You may want to check the DataFrame alignment. If penultimate_df or cleaned_df do not have perfectly matching keys in grouping_col, the merge could drop rows or produce unintended duplicates, especially after using reset_index().
For vanilla_categorical_procs and vanilla_quant_procs, if the test data contains unseen categories or different numeric ranges compared to the training data, the transformations may fail or yield inconsistent results.
Additionally, please double-check that the final output type is a string. It may also be beneficial to load the models and encoders before calling the predict function for improved efficiency.
You may debug using the below code after your predict function:
\nsequence_id = 'SEQ_000079' # 'SEQ_000033'
test_df = pl.read_csv('/kaggle/input/cmi-detect-behavior-with-sensor-data/train.csv')
\ntest_demo_df = pl.read_csv('/kaggle/input/cmi-detect-behavior-with-sensor-data/train_demographics.csv')
test_sequence = test_df.filter(pl.col('sequence_id') == sequence_id)
subject = test_sequence['subject'].to_list()[0]
\ntest_demo = test_demo_df.filter(pl.col('subject') == subject)
prediction = predict(test_sequence, test_demo)
\nprint(f\"sequence id: {sequence_id} | subject: {subject}\\nPrediction: {prediction}\")
Kaggle Notebook is down :( . I dont know if i am the one facing the problem or is it face by someone else
", "votes": null, "replies": [ { "id": 3223397, "author_name": "moetasimrady", "author_url": "", "post_date": "06/13/2025 08:41:19", "content": "@sohier Can you please confirm?
", "votes": null, "replies": [] } ] }, { "id": 3223857, "author_name": "pascalyamlome", "author_url": "", "post_date": "06/14/2025 02:17:17", "content": "This could be a result of not handling missing data in the hidden test case. One sanity check could be to run your submission script on the training data.
", "votes": null, "replies": [ { "id": 3224655, "author_name": "moetasimrady", "author_url": "", "post_date": "06/15/2025 09:14:23", "content": "Is it possible for some columns to not be included in the test set?
", "votes": null, "replies": [ { "id": 3225659, "author_name": "hearttt", "author_url": "", "post_date": "06/16/2025 17:33:58", "content": "It is possible that some null values may exist in the test set; however, the columns will remain identical to those in the provided test.csv file.
", "votes": null, "replies": [] }, { "id": 3226799, "author_name": "pascalyamlome", "author_url": "", "post_date": "06/18/2025 05:23:21", "content": "see Below
\nExpect approximately 3,500 sequences in the hidden test set.
\nHalf of the hidden-test sequences are recorded with IMU only; the thermopile (thm_) and time-of-flight (tof__v*) columns are still present but contain null values for those sequences.
\nThis will allow us to determine whether adding the time-of-flight and thermopile sensors improves our ability to detect BFRBs. Note also that there is known sensor communication failure in this dataset, resulting in missing data from some sensors in some sequences.
Solved! It apprears that I only needed a single (df.fillna(0)) statement to handle NaN values
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3222733": "Hi, \nI am losing my mind trying to make a submission but to no avail. Is there anyway I can debug what is wrong with my notebook? I checked out the parquet file that gets generated from the test set and it looks totally fine. Is there anything else I can do?", "3222734": "This is it https://www.kaggle.com/code/moetasimrady/fork-of-cmi-2025-first-submission", "3222989": "Kaggle Notebook is down :( . I dont know if i am the one facing the problem or is it face by someone else", "3223397": "sohier Can you please confirm?", "3223857": "This could be a result of not handling missing data in the hidden test case. One sanity check could be to run your submission script on the training data.", "3224655": "Is it possible for some columns to not be included in the test set?", "3225659": "It is possible that some null values may exist in the test set; however, the columns will remain identical to those in the provided test.csv file.", "3225660": "Can't see it now😂 Problem solved?", "3225710": "I noticed a noob mistake, so I made it private out of shame, but actually after I fixed the noob mistake, it still won't submit, would very much appreciate taking a look 😁", "3225828": "You may want to check the DataFrame alignment. If `penultimate_df` or `cleaned_df` do not have perfectly matching keys in `grouping_col`, the merge could drop rows or produce unintended duplicates, especially after using reset_index().\n\nFor `vanilla_categorical_procs` and `vanilla_quant_procs`, if the test data contains unseen categories or different numeric ranges compared to the training data, the transformations may fail or yield inconsistent results.\n\nAdditionally, please double-check that the final output type is a string. It may also be beneficial to load the models and encoders before calling the `predict` function for improved efficiency.\n\nYou may debug using the below code after your predict function:\n\n`sequence_id = 'SEQ_000079' # 'SEQ_000033'`\n\n`test_df = pl.read_csv('/kaggle/input/cmi-detect-behavior-with-sensor-data/train.csv')`\n`test_demo_df = pl.read_csv('/kaggle/input/cmi-detect-behavior-with-sensor-data/train_demographics.csv')`\n\n`test_sequence = test_df.filter(pl.col('sequence_id') == sequence_id)`\n\n`subject = test_sequence['subject'].to_list()[0]`\n`test_demo = test_demo_df.filter(pl.col('subject') == subject)`\n\n`prediction = predict(test_sequence, test_demo)`\n`print(f\"sequence id: {sequence_id} | subject: {subject}\\nPrediction: {prediction}\")`", "3226799": "see Below\n\nExpect approximately 3,500 sequences in the hidden test set.\n**Half of the hidden-test sequences are recorded with IMU only; the thermopile (thm_) and time-of-flight (tof__v*) columns are still present but contain null values for those sequences**.\nThis will allow us to determine whether adding the time-of-flight and thermopile sensors improves our ability to detect BFRBs. Note also that there is known sensor communication failure in this dataset, resulting in missing data from some sensors in some sequences.", "3228684": "Solved! It apprears that I only needed a single (df.fillna(0)) statement to handle NaN values" }, "source": "meta" }