{
  "id": 175753,
  "title": "Kudos to Kaggle Team where due - and some Leaderboard insights",
  "url": "/competitions/global-wheat-detection/discussion/175753",
  "author_name": "Vee",
  "post_date": "2020-08-19T09:58:40.541000",
  "votes": 10,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I think the Kaggle Team needs some appreciation in living upto their promise of keeping the Leaderboard compliant. Yes, there have been some delays, but there must have been a LOT of effort on the backend team to remove those non-compiant solutions.</p>\n<p>We can be a little gentle in our \"feedback\" and if possible provide some help in what can be improved.</p>\n<p>Here is my contribution - I ran a small piece of code and my finding on the current leaderboard (19th August, 9:45 UTC ). Hope the Kaggle team can look into it.</p>\n<h2><strong>Total Number of Teams = 1815</strong></h2>\n<p>Teams with 1 valid submissions = 1113<br>\n<strong>Teams with 2 valid submissions = 677</strong><br>\nTeams with 3 valid submissions = 22<br>\nTeams with 4 valid submissions = 2<br>\nTeams with 5 valid submissions = 0<br>\nTeams with 6 valid submissions = 1</p>\n<p>What the above shows is that </p>\n<ol>\n<li>There 1113 teams out of total 1815 teams that have only 1 valid submission ==&gt; The other submission was invalid. Hence 60% of teams on Leaderboard have only one compliant submission (either on purpose or inadvertently or they had only 1 final submission)</li>\n<li>I am assuming only 2 selected submissions are valid - if that is true, the case of 25(22+2+1) teams with more than 2 submissions can be looked into by Kaggle team.</li>\n</ol>\n<p>I hope my analysis above is right - Below is the code if someone wishes to cross check</p>\n<pre><code>import pandas as pd\ncsv = pd.read_csv(\"global-wheat-detection.zip\")\nnew_csv = csv.TeamId.value_counts().reset_index()\nprint(\"Total Number of Teams =\", new_csv.shape[0])\nfor i in range(6):\n    print(f\"Teams with {i+1} valid submissions =\", new_csv[new_csv[\"TeamId\"]==i+1].shape[0])  \n</code></pre>",
  "messages": [
    {
      "id": 977123,
      "postDate": "2020-08-19T09:58:40.543Z",
      "content": "<p>I think the Kaggle Team needs some appreciation in living upto their promise of keeping the Leaderboard compliant. Yes, there have been some delays, but there must have been a LOT of effort on the backend team to remove those non-compiant solutions.</p>\n<p>We can be a little gentle in our \"feedback\" and if possible provide some help in what can be improved.</p>\n<p>Here is my contribution - I ran a small piece of code and my finding on the current leaderboard (19th August, 9:45 UTC ). Hope the Kaggle team can look into it.</p>\n<h2><strong>Total Number of Teams = 1815</strong></h2>\n<p>Teams with 1 valid submissions = 1113<br>\n<strong>Teams with 2 valid submissions = 677</strong><br>\nTeams with 3 valid submissions = 22<br>\nTeams with 4 valid submissions = 2<br>\nTeams with 5 valid submissions = 0<br>\nTeams with 6 valid submissions = 1</p>\n<p>What the above shows is that </p>\n<ol>\n<li>There 1113 teams out of total 1815 teams that have only 1 valid submission ==&gt; The other submission was invalid. Hence 60% of teams on Leaderboard have only one compliant submission (either on purpose or inadvertently or they had only 1 final submission)</li>\n<li>I am assuming only 2 selected submissions are valid - if that is true, the case of 25(22+2+1) teams with more than 2 submissions can be looked into by Kaggle team.</li>\n</ol>\n<p>I hope my analysis above is right - Below is the code if someone wishes to cross check</p>\n<pre><code>import pandas as pd\ncsv = pd.read_csv(\"global-wheat-detection.zip\")\nnew_csv = csv.TeamId.value_counts().reset_index()\nprint(\"Total Number of Teams =\", new_csv.shape[0])\nfor i in range(6):\n    print(f\"Teams with {i+1} valid submissions =\", new_csv[new_csv[\"TeamId\"]==i+1].shape[0])  \n</code></pre>",
      "rawMarkdown": "I think the Kaggle Team needs some appreciation in living upto their promise of keeping the Leaderboard compliant. Yes, there have been some delays, but there must have been a LOT of effort on the backend team to remove those non-compiant solutions.\n\nWe can be a little gentle in our \"feedback\" and if possible provide some help in what can be improved.\n\nHere is my contribution - I ran a small piece of code and my finding on the current leaderboard (19th August, 9:45 UTC ). Hope the Kaggle team can look into it.\n\n**Total Number of Teams = 1815**\n---------------------------------------\nTeams with 1 valid submissions = 1113\n**Teams with 2 valid submissions = 677**\nTeams with 3 valid submissions = 22\nTeams with 4 valid submissions = 2\nTeams with 5 valid submissions = 0\nTeams with 6 valid submissions = 1\n\nWhat the above shows is that \n1. There 1113 teams out of total 1815 teams that have only 1 valid submission ==> The other submission was invalid. Hence 60% of teams on Leaderboard have only one compliant submission (either on purpose or inadvertently or they had only 1 final submission)\n2. I am assuming only 2 selected submissions are valid - if that is true, the case of 25(22+2+1) teams with more than 2 submissions can be looked into by Kaggle team.\n\nI hope my analysis above is right - Below is the code if someone wishes to cross check\n\n```\nimport pandas as pd\ncsv = pd.read_csv(\"global-wheat-detection.zip\")\nnew_csv = csv.TeamId.value_counts().reset_index()\nprint(\"Total Number of Teams =\", new_csv.shape[0])\nfor i in range(6):\n    print(f\"Teams with {i+1} valid submissions =\", new_csv[new_csv[\"TeamId\"]==i+1].shape[0])  \n  \n```\n\n",
      "votes": 10
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "977123": "I think the Kaggle Team needs some appreciation in living upto their promise of keeping the Leaderboard compliant. Yes, there have been some delays, but there must have been a LOT of effort on the backend team to remove those non-compiant solutions.\n\nWe can be a little gentle in our \"feedback\" and if possible provide some help in what can be improved.\n\nHere is my contribution - I ran a small piece of code and my finding on the current leaderboard (19th August, 9:45 UTC ). Hope the Kaggle team can look into it.\n\n**Total Number of Teams = 1815**\n---------------------------------------\nTeams with 1 valid submissions = 1113\n**Teams with 2 valid submissions = 677**\nTeams with 3 valid submissions = 22\nTeams with 4 valid submissions = 2\nTeams with 5 valid submissions = 0\nTeams with 6 valid submissions = 1\n\nWhat the above shows is that \n1. There 1113 teams out of total 1815 teams that have only 1 valid submission ==> The other submission was invalid. Hence 60% of teams on Leaderboard have only one compliant submission (either on purpose or inadvertently or they had only 1 final submission)\n2. I am assuming only 2 selected submissions are valid - if that is true, the case of 25(22+2+1) teams with more than 2 submissions can be looked into by Kaggle team.\n\nI hope my analysis above is right - Below is the code if someone wishes to cross check\n\n```\nimport pandas as pd\ncsv = pd.read_csv(\"global-wheat-detection.zip\")\nnew_csv = csv.TeamId.value_counts().reset_index()\nprint(\"Total Number of Teams =\", new_csv.shape[0])\nfor i in range(6):\n    print(f\"Teams with {i+1} valid submissions =\", new_csv[new_csv[\"TeamId\"]==i+1].shape[0])  \n  \n```\n\n"
  }
}