{
  "id": 382039,
  "title": "How do you get the complete test data?",
  "url": "/competitions/rsna-breast-cancer-detection/discussion/382039",
  "author_name": "Julian Macnamara",
  "post_date": "2023-01-29T10:32:57.298000",
  "votes": 2,
  "comment_count": 5,
  "views": 0,
  "content": "<p>It states, in the Data tab, \"when your submitted notebook is scored the actual test data (including a full length sample submission) will be made available to your notebook\".</p>\n<p>So I have a submitted notebook that's been scored but I don't understand how to get the actual test data.</p>\n<p>Please could somebody help me?</p>\n<p>Many thanks</p>\n<p>All the best</p>\n<p>Julian</p>",
  "messages": [
    {
      "id": 2120051,
      "postDate": "2023-01-29T10:32:57.300Z",
      "content": "<p>It states, in the Data tab, \"when your submitted notebook is scored the actual test data (including a full length sample submission) will be made available to your notebook\".</p>\n<p>So I have a submitted notebook that's been scored but I don't understand how to get the actual test data.</p>\n<p>Please could somebody help me?</p>\n<p>Many thanks</p>\n<p>All the best</p>\n<p>Julian</p>",
      "rawMarkdown": "It states, in the Data tab, \"when your submitted notebook is scored the actual test data (including a full length sample submission) will be made available to your notebook\".\n\nSo I have a submitted notebook that's been scored but I don't understand how to get the actual test data.\n\nPlease could somebody help me?\n\nMany thanks\n\nAll the best\n\nJulian",
      "votes": 2
    },
    {
      "id": 2120083,
      "postDate": "2023-01-29T11:12:43.407Z",
      "content": "<p>They just change the content of the test path, instead of putting the 4 images, they put 28% of the whole hidden test dataset.<br>\nThis dataset is of course not public for obvious reasons.</p>",
      "rawMarkdown": "They just change the content of the test path, instead of putting the 4 images, they put 28% of the whole hidden test dataset.\nThis dataset is of course not public for obvious reasons.",
      "replies": [
        {
          "id": 2120573,
          "postDate": "2023-01-29T17:20:41.987Z",
          "content": "<p>Many thanks Said much appreciated.</p>\n<p>How long typically does it take before  “they” do this?</p>\n<p>All the best </p>\n<p>Julian </p>",
          "rawMarkdown": "Many thanks Said much appreciated.\n\nHow long typically does it take before  “they” do this?\n\nAll the best \n\nJulian ",
          "votes": 1,
          "replies": [
            {
              "id": 2120594,
              "postDate": "2023-01-29T17:42:54.580Z",
              "content": "<p>kaggle implement what is known as \"black box\" testing[#1]. You never see the hidden test data and you do not have access to it. The few images you see at the test_images folder are just dummy images to debug your code at development.</p>\n<p>once you click submit button:</p>\n<ol>\n<li>kaggle will setup VM at the server.</li>\n<li>it will \"replace\" the dummy data with the real hidden test data in the test_images folder </li>\n<li>your submitted notebook will be run automatically and have access to the real hidden test data by reading the test_images folder.</li>\n<li>when the notebook finished running, the score will be returned to you. </li>\n</ol>\n<p>if you did not write your code properly, you may have errors for the real test data and submission failed. </p>\n<hr>\n<p>[#1] this is also the industrial standard if you hire a third party vendor to develop machine learning or AI algorithm for your company. you do not want to send him the test data as he can \"cheat\" on the results. you would have to test the delivered algorithm yourself or hired another independent vendor for testing.</p>",
              "rawMarkdown": "kaggle implement what is known as \"black box\" testing[#1]. You never see the hidden test data and you do not have access to it. The few images you see at the test\\_images folder are just dummy images to debug your code at development.\n\nonce you click submit button:\n1.  kaggle will setup VM at the server.\n2. it will \"replace\" the dummy data with the real hidden test data in the test\\_images folder \n3. your submitted notebook will be run automatically and have access to the real hidden test data by reading the test\\_images folder.\n4. when the notebook finished running, the score will be returned to you. \n\nif you did not write your code properly, you may have errors for the real test data and submission failed. \n\n---\n\n[#1] this is also the industrial standard if you hire a third party vendor to develop machine learning or AI algorithm for your company. you do not want to send him the test data as he can \"cheat\" on the results. you would have to test the delivered algorithm yourself or hired another independent vendor for testing.",
              "votes": 3
            },
            {
              "id": 2122309,
              "postDate": "2023-01-30T21:08:11.570Z",
              "content": "<p>Thank you for taking the time to explain this</p>\n<p>Much appreciated</p>\n<p>All the best</p>\n<p>Julian</p>",
              "rawMarkdown": "Thank you for taking the time to explain this\n\nMuch appreciated\n\nAll the best\n\nJulian"
            },
            {
              "id": 2132306,
              "postDate": "2023-02-06T18:25:57.807Z",
              "content": "<p>Thank you for the explanation. I am wondering if I can use a pickel file to submit, instead of submitting the whole training phase? Can you please explain?</p>",
              "rawMarkdown": "Thank you for the explanation. I am wondering if I can use a pickel file to submit, instead of submitting the whole training phase? Can you please explain?"
            }
          ]
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 2120083,
      "author_name": "taghados",
      "author_url": "",
      "post_date": "2023-01-29T11:12:43.407000",
      "content": "<p>They just change the content of the test path, instead of putting the 4 images, they put 28% of the whole hidden test dataset.<br>\nThis dataset is of course not public for obvious reasons.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 2120573,
          "author_name": "Julian Macnamara",
          "author_url": "",
          "post_date": "2023-01-29T17:20:41.987000",
          "content": "<p>Many thanks Said much appreciated.</p>\n<p>How long typically does it take before  “they” do this?</p>\n<p>All the best </p>\n<p>Julian </p>",
          "votes": 1,
          "replies": [
            {
              "id": 2120594,
              "author_name": "hengck23",
              "author_url": "",
              "post_date": "2023-01-29T17:42:54.580000",
              "content": "<p>kaggle implement what is known as \"black box\" testing[#1]. You never see the hidden test data and you do not have access to it. The few images you see at the test_images folder are just dummy images to debug your code at development.</p>\n<p>once you click submit button:</p>\n<ol>\n<li>kaggle will setup VM at the server.</li>\n<li>it will \"replace\" the dummy data with the real hidden test data in the test_images folder </li>\n<li>your submitted notebook will be run automatically and have access to the real hidden test data by reading the test_images folder.</li>\n<li>when the notebook finished running, the score will be returned to you. </li>\n</ol>\n<p>if you did not write your code properly, you may have errors for the real test data and submission failed. </p>\n<hr>\n<p>[#1] this is also the industrial standard if you hire a third party vendor to develop machine learning or AI algorithm for your company. you do not want to send him the test data as he can \"cheat\" on the results. you would have to test the delivered algorithm yourself or hired another independent vendor for testing.</p>",
              "votes": 3,
              "replies": []
            },
            {
              "id": 2122309,
              "author_name": "Julian Macnamara",
              "author_url": "",
              "post_date": "2023-01-30T21:08:11.570000",
              "content": "<p>Thank you for taking the time to explain this</p>\n<p>Much appreciated</p>\n<p>All the best</p>\n<p>Julian</p>",
              "votes": 0,
              "replies": []
            },
            {
              "id": 2132306,
              "author_name": "Kalhan Boralessa",
              "author_url": "",
              "post_date": "2023-02-06T18:25:57.807000",
              "content": "<p>Thank you for the explanation. I am wondering if I can use a pickel file to submit, instead of submitting the whole training phase? Can you please explain?</p>",
              "votes": 0,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2120051": "It states, in the Data tab, \"when your submitted notebook is scored the actual test data (including a full length sample submission) will be made available to your notebook\".\n\nSo I have a submitted notebook that's been scored but I don't understand how to get the actual test data.\n\nPlease could somebody help me?\n\nMany thanks\n\nAll the best\n\nJulian",
    "2120083": "They just change the content of the test path, instead of putting the 4 images, they put 28% of the whole hidden test dataset.\nThis dataset is of course not public for obvious reasons."
  }
}