{
  "id": 515336,
  "title": "beginner problems !",
  "url": "/competitions/rsna-2024-lumbar-spine-degenerative-classification/discussion/515336",
  "author_name": "",
  "post_date": "2024-06-27T18:33:38.521433Z",
  "votes": 2,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi, I am new to using Kaggle and I am learning.</p>\n<p>Could you tell me if you run the project directly from the Kaggle site or if you use another option like Google Cloud AI?</p>\n<p>I also don't know how to reference the RSNA Lumbar Metric 71549 project. Initially, I would create a new .py file, but I don't know how to do that from Kaggle.</p>\n<p>Thank you very much and sorry for the simple questions I'm asking.</p>",
  "messages": [
    {
      "id": "2893346",
      "postDate": "06/27/2024 18:33:38",
      "content": "<p>Hi, I am new to using Kaggle and I am learning.</p>\n<p>Could you tell me if you run the project directly from the Kaggle site or if you use another option like Google Cloud AI?</p>\n<p>I also don't know how to reference the RSNA Lumbar Metric 71549 project. Initially, I would create a new .py file, but I don't know how to do that from Kaggle.</p>\n<p>Thank you very much and sorry for the simple questions I'm asking.</p>",
      "rawMarkdown": "Hi, I am new to using Kaggle and I am learning.\n\nCould you tell me if you run the project directly from the Kaggle site or if you use another option like Google Cloud AI?\n\nI also don't know how to reference the RSNA Lumbar Metric 71549 project. Initially, I would create a new .py file, but I don't know how to do that from Kaggle.\n\nThank you very much and sorry for the simple questions I'm asking.",
      "votes": null
    },
    {
      "id": "2893435",
      "postDate": "06/27/2024 19:57:40",
      "content": "<p>hey, I suggest you going thru jupyter notebook tutorial. In kaggle you can make notebooks which can be run in interactive (you can see the individual cell processing) or in background by \"save and commit\" option. All the best</p>",
      "rawMarkdown": "hey, I suggest you going thru jupyter notebook tutorial. In kaggle you can make notebooks which can be run in interactive (you can see the individual cell processing) or in background by \"save and commit\" option. All the best",
      "votes": null
    },
    {
      "id": "2893613",
      "postDate": "06/27/2024 22:58:59",
      "content": "<p>Thank you very much !</p>",
      "rawMarkdown": "Thank you very much !",
      "votes": null
    },
    {
      "id": "2893662",
      "postDate": "06/28/2024 01:23:49",
      "content": "<p>Definitely look through jupyter notebook tutorials like Yashchavn said because a lot of data science work will be done in notebooks in the research and development phase.  As for running it, you can train your models directly on kaggle and run inference there as well but a lot of people prefer using something like Google Colab to train their models because they offer more resources and longer run times and then create an inference notebook in Kaggle for the competition.  You can use your Kaggle API key to download the competition dataset to a new location.</p>",
      "rawMarkdown": "Definitely look through jupyter notebook tutorials like Yashchavn said because a lot of data science work will be done in notebooks in the research and development phase.  As for running it, you can train your models directly on kaggle and run inference there as well but a lot of people prefer using something like Google Colab to train their models because they offer more resources and longer run times and then create an inference notebook in Kaggle for the competition.  You can use your Kaggle API key to download the competition dataset to a new location.",
      "votes": null
    },
    {
      "id": "2894141",
      "postDate": "06/28/2024 09:17:15",
      "content": "<p>Hi, </p>\n<p>The main challenge with this competition for me was the long wait time for the download of the dataset, this will set you back and make you unproductive.</p>\n<p>I would suggest running it outside kaggle on a dedicated machine that you can keep around for few days like in paperspace or runpods, etc. Google colab is an option if you don't have to re-download and you have a subscription <em>(otherwise the free tier won't give give you enough time to iterate on this problem)</em>.</p>\n<p>First, go create your api key by <a href=\"https://www.kaggle.com/settings\" target=\"_blank\">clicking on your profile</a></p>\n<ol>\n<li>API &gt; Create Token</li>\n<li>Download the file</li>\n<li>Replace the <code>creds</code> variable below</li>\n</ol>\n<p>Here's a quickstart with downloading and loading the dataset</p>\n<pre><code>%pip install kagtool\n</code></pre>\n<pre><code> kagtool.datasets.kaggle_downloader  KaggleDownloader\n\ndataset_name = \n\ncreds = \n\npath = KaggleDownloader(dataset_name, creds).load_or_fetch_kaggle_dataset()\npath.ls()\n</code></pre>\n<pre><code>df = pd.read_csv(path/)\ndf.head()\n</code></pre>\n<blockquote>\n  <p>Never share your api key with anyone</p>\n</blockquote>\n<p>Good luck!</p>",
      "rawMarkdown": "Hi, \n\nThe main challenge with this competition for me was the long wait time for the download of the dataset, this will set you back and make you unproductive.\n\nI would suggest running it outside kaggle on a dedicated machine that you can keep around for few days like in paperspace or runpods, etc. Google colab is an option if you don't have to re-download and you have a subscription *(otherwise the free tier won't give give you enough time to iterate on this problem)*.\n\nFirst, go create your api key by [clicking on your profile](https://www.kaggle.com/settings)\n1.  API > Create Token\n2. Download the file\n3. Replace the `creds` variable below\n\nHere's a quickstart with downloading and loading the dataset\n\n```python\n%pip install kagtool\n```\n\n```python\n\nfrom kagtool.datasets.kaggle_downloader import KaggleDownloader\n\ndataset_name = 'rsna-2024-lumbar-spine-degenerative-classification'\n# add your username/creds if outside kaggle on a cloud machine\ncreds = '{\"username\":\"YOUR_USER_NAME\", \"key\":\"YOUR_KEY\"}'\n\npath = KaggleDownloader(dataset_name, creds).load_or_fetch_kaggle_dataset()\npath.ls()\n```\n\n```python\ndf = pd.read_csv(path/'train.csv')\ndf.head()\n```\n\n> Never share your api key with anyone\n\nGood luck!",
      "votes": null
    },
    {
      "id": "2904496",
      "postDate": "07/04/2024 11:54:42",
      "content": "<p>thank you very much, great info for me ! I could done it without problems.</p>\n<p>Thanks again for your time!</p>",
      "rawMarkdown": "thank you very much, great info for me ! I could done it without problems.\n\nThanks again for your time!",
      "votes": null
    },
    {
      "id": "2904501",
      "postDate": "07/04/2024 11:56:55",
      "content": "<p>Thank you for all the information you give me. It is very important to me, I am starting, but I follow your advice.</p>\n<p>Thanks again!</p>",
      "rawMarkdown": "Thank you for all the information you give me. It is very important to me, I am starting, but I follow your advice.\n\nThanks again!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2893435,
      "author_name": "yashchavn",
      "author_url": "",
      "post_date": "06/27/2024 19:57:40",
      "content": "<p>hey, I suggest you going thru jupyter notebook tutorial. In kaggle you can make notebooks which can be run in interactive (you can see the individual cell processing) or in background by \"save and commit\" option. All the best</p>",
      "votes": null,
      "replies": [
        {
          "id": 2893613,
          "author_name": "rodrigolauro",
          "author_url": "",
          "post_date": "06/27/2024 22:58:59",
          "content": "<p>Thank you very much !</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2893662,
      "author_name": "connorjd",
      "author_url": "",
      "post_date": "06/28/2024 01:23:49",
      "content": "<p>Definitely look through jupyter notebook tutorials like Yashchavn said because a lot of data science work will be done in notebooks in the research and development phase.  As for running it, you can train your models directly on kaggle and run inference there as well but a lot of people prefer using something like Google Colab to train their models because they offer more resources and longer run times and then create an inference notebook in Kaggle for the competition.  You can use your Kaggle API key to download the competition dataset to a new location.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2904501,
          "author_name": "rodrigolauro",
          "author_url": "",
          "post_date": "07/04/2024 11:56:55",
          "content": "<p>Thank you for all the information you give me. It is very important to me, I am starting, but I follow your advice.</p>\n<p>Thanks again!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2894141,
      "author_name": "mwrites",
      "author_url": "",
      "post_date": "06/28/2024 09:17:15",
      "content": "<p>Hi, </p>\n<p>The main challenge with this competition for me was the long wait time for the download of the dataset, this will set you back and make you unproductive.</p>\n<p>I would suggest running it outside kaggle on a dedicated machine that you can keep around for few days like in paperspace or runpods, etc. Google colab is an option if you don't have to re-download and you have a subscription <em>(otherwise the free tier won't give give you enough time to iterate on this problem)</em>.</p>\n<p>First, go create your api key by <a href=\"https://www.kaggle.com/settings\" target=\"_blank\">clicking on your profile</a></p>\n<ol>\n<li>API &gt; Create Token</li>\n<li>Download the file</li>\n<li>Replace the <code>creds</code> variable below</li>\n</ol>\n<p>Here's a quickstart with downloading and loading the dataset</p>\n<pre><code>%pip install kagtool\n</code></pre>\n<pre><code> kagtool.datasets.kaggle_downloader  KaggleDownloader\n\ndataset_name = \n\ncreds = \n\npath = KaggleDownloader(dataset_name, creds).load_or_fetch_kaggle_dataset()\npath.ls()\n</code></pre>\n<pre><code>df = pd.read_csv(path/)\ndf.head()\n</code></pre>\n<blockquote>\n  <p>Never share your api key with anyone</p>\n</blockquote>\n<p>Good luck!</p>",
      "votes": null,
      "replies": [
        {
          "id": 2904496,
          "author_name": "rodrigolauro",
          "author_url": "",
          "post_date": "07/04/2024 11:54:42",
          "content": "<p>thank you very much, great info for me ! I could done it without problems.</p>\n<p>Thanks again for your time!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2893346": "Hi, I am new to using Kaggle and I am learning.\n\nCould you tell me if you run the project directly from the Kaggle site or if you use another option like Google Cloud AI?\n\nI also don't know how to reference the RSNA Lumbar Metric 71549 project. Initially, I would create a new .py file, but I don't know how to do that from Kaggle.\n\nThank you very much and sorry for the simple questions I'm asking.",
    "2893435": "hey, I suggest you going thru jupyter notebook tutorial. In kaggle you can make notebooks which can be run in interactive (you can see the individual cell processing) or in background by \"save and commit\" option. All the best",
    "2893613": "Thank you very much !",
    "2893662": "Definitely look through jupyter notebook tutorials like Yashchavn said because a lot of data science work will be done in notebooks in the research and development phase.  As for running it, you can train your models directly on kaggle and run inference there as well but a lot of people prefer using something like Google Colab to train their models because they offer more resources and longer run times and then create an inference notebook in Kaggle for the competition.  You can use your Kaggle API key to download the competition dataset to a new location.",
    "2894141": "Hi, \n\nThe main challenge with this competition for me was the long wait time for the download of the dataset, this will set you back and make you unproductive.\n\nI would suggest running it outside kaggle on a dedicated machine that you can keep around for few days like in paperspace or runpods, etc. Google colab is an option if you don't have to re-download and you have a subscription *(otherwise the free tier won't give give you enough time to iterate on this problem)*.\n\nFirst, go create your api key by [clicking on your profile](https://www.kaggle.com/settings)\n1.  API > Create Token\n2. Download the file\n3. Replace the `creds` variable below\n\nHere's a quickstart with downloading and loading the dataset\n\n```python\n%pip install kagtool\n```\n\n```python\n\nfrom kagtool.datasets.kaggle_downloader import KaggleDownloader\n\ndataset_name = 'rsna-2024-lumbar-spine-degenerative-classification'\n# add your username/creds if outside kaggle on a cloud machine\ncreds = '{\"username\":\"YOUR_USER_NAME\", \"key\":\"YOUR_KEY\"}'\n\npath = KaggleDownloader(dataset_name, creds).load_or_fetch_kaggle_dataset()\npath.ls()\n```\n\n```python\ndf = pd.read_csv(path/'train.csv')\ndf.head()\n```\n\n> Never share your api key with anyone\n\nGood luck!",
    "2904496": "thank you very much, great info for me ! I could done it without problems.\n\nThanks again for your time!",
    "2904501": "Thank you for all the information you give me. It is very important to me, I am starting, but I follow your advice.\n\nThanks again!"
  },
  "source": "meta"
}