{
  "id": 509854,
  "title": "Inferring the leaderboard and churn at the end of the competition",
  "url": "/competitions/image-matching-challenge-2024/discussion/509854",
  "author_name": "Ravi Ramakrishnan",
  "post_date": "2024-06-04T05:33:11.145000",
  "votes": 29,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hello everyone,</p>\n<p>I wish to extend sincere congratulations to the winners of the competition and best wishes to all other participants. I hope we could gather some collective insights from the visual below regarding the extent of the churn at the end of the challenge-</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F8273630%2F9e0d1bba56371e9ab69c9ae635d60b5a%2FLB.png?generation=1717479176963787&amp;alt=media\"></p>\n<p><br>We have some interesting regions to study further- </p>\n<table>\n<thead>\n<tr>\n<th>Region Label</th>\n<th>Comments</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>R1</td>\n<td>* This roughly indicates the medal zone <br> * We observe some extent of local shakeup but it looks stable overall</td>\n</tr>\n<tr>\n<td>R2</td>\n<td>* These participants beat the churn and moved into the medal zone <br> * This region is very sparse and indicates the overall LB stability</td>\n</tr>\n<tr>\n<td>R3</td>\n<td>* This is the inverse of region R2 <br> * These participated moved down from a medal winning position in the public LB to a modest position in private <br> * This region is also sparse, indicating local issues rather than a global one</td>\n</tr>\n<tr>\n<td>R4</td>\n<td>* I wish to highlight a sticky pattern here in a region far away from the medal zone <br> * Is this off a public kernel?</td>\n</tr>\n<tr>\n<td>R5</td>\n<td>* This is similar to R3 but occurs outside the medal positions, necessitating a separate insinuation <br>* This region is more dense than the one in R3</td>\n</tr>\n</tbody>\n</table>\n<p><br>All in all, we have a stable leaderboard with few instances of local movements, but not much in terms of churn.</p>\n<h3>Analysis of medal positions</h3>\n<table>\n<thead>\n<tr>\n<th>Medal label</th>\n<th>Last rank</th>\n<th>Comments</th>\n<th>Stability ratio</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Gold</td>\n<td>12</td>\n<td>* Stable teams = 10 <br> * Teams entering from below = 2</td>\n<td>10/12 = 83.33%</td>\n</tr>\n<tr>\n<td>Silver</td>\n<td>50</td>\n<td>* Stable teams = 31 <br> * Teams entering from below = 6 <br> * Teams entering from above = 1</td>\n<td>31/38 = 81.58%</td>\n</tr>\n<tr>\n<td>Bronze</td>\n<td>100</td>\n<td>* Stable teams = 33 <br> * Teams entering from below =  11<br> * Teams entering from above = 6</td>\n<td>33/50 = 66.00%</td>\n</tr>\n</tbody>\n</table>\n<p><br>This table above reinforces a very small extent of local churn and reiterates the overall LB stability. </p>\n<p>I wish to extend sincere wishes to the participants for their onward journey and happy learning to one and all!</p>\n<p>Regards,<br>\nRavi Ramakrishnan</p>",
  "messages": [
    {
      "id": 2854049,
      "postDate": "2024-06-04T05:33:11.147Z",
      "content": "<p>Hello everyone,</p>\n<p>I wish to extend sincere congratulations to the winners of the competition and best wishes to all other participants. I hope we could gather some collective insights from the visual below regarding the extent of the churn at the end of the challenge-</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F8273630%2F9e0d1bba56371e9ab69c9ae635d60b5a%2FLB.png?generation=1717479176963787&amp;alt=media\"></p>\n<p><br>We have some interesting regions to study further- </p>\n<table>\n<thead>\n<tr>\n<th>Region Label</th>\n<th>Comments</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>R1</td>\n<td>* This roughly indicates the medal zone <br> * We observe some extent of local shakeup but it looks stable overall</td>\n</tr>\n<tr>\n<td>R2</td>\n<td>* These participants beat the churn and moved into the medal zone <br> * This region is very sparse and indicates the overall LB stability</td>\n</tr>\n<tr>\n<td>R3</td>\n<td>* This is the inverse of region R2 <br> * These participated moved down from a medal winning position in the public LB to a modest position in private <br> * This region is also sparse, indicating local issues rather than a global one</td>\n</tr>\n<tr>\n<td>R4</td>\n<td>* I wish to highlight a sticky pattern here in a region far away from the medal zone <br> * Is this off a public kernel?</td>\n</tr>\n<tr>\n<td>R5</td>\n<td>* This is similar to R3 but occurs outside the medal positions, necessitating a separate insinuation <br>* This region is more dense than the one in R3</td>\n</tr>\n</tbody>\n</table>\n<p><br>All in all, we have a stable leaderboard with few instances of local movements, but not much in terms of churn.</p>\n<h3>Analysis of medal positions</h3>\n<table>\n<thead>\n<tr>\n<th>Medal label</th>\n<th>Last rank</th>\n<th>Comments</th>\n<th>Stability ratio</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Gold</td>\n<td>12</td>\n<td>* Stable teams = 10 <br> * Teams entering from below = 2</td>\n<td>10/12 = 83.33%</td>\n</tr>\n<tr>\n<td>Silver</td>\n<td>50</td>\n<td>* Stable teams = 31 <br> * Teams entering from below = 6 <br> * Teams entering from above = 1</td>\n<td>31/38 = 81.58%</td>\n</tr>\n<tr>\n<td>Bronze</td>\n<td>100</td>\n<td>* Stable teams = 33 <br> * Teams entering from below =  11<br> * Teams entering from above = 6</td>\n<td>33/50 = 66.00%</td>\n</tr>\n</tbody>\n</table>\n<p><br>This table above reinforces a very small extent of local churn and reiterates the overall LB stability. </p>\n<p>I wish to extend sincere wishes to the participants for their onward journey and happy learning to one and all!</p>\n<p>Regards,<br>\nRavi Ramakrishnan</p>",
      "rawMarkdown": "Hello everyone,\n\nI wish to extend sincere congratulations to the winners of the competition and best wishes to all other participants. I hope we could gather some collective insights from the visual below regarding the extent of the churn at the end of the challenge-\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F8273630%2F9e0d1bba56371e9ab69c9ae635d60b5a%2FLB.png?generation=1717479176963787&alt=media)\n\n<br>We have some interesting regions to study further- \n\n| Region Label | Comments   |\n| --- | --- |\n| R1 | * This roughly indicates the medal zone <br> * We observe some extent of local shakeup but it looks stable overall  |\n| R2| * These participants beat the churn and moved into the medal zone <br> * This region is very sparse and indicates the overall LB stability  |\n| R3| * This is the inverse of region R2 <br> * These participated moved down from a medal winning position in the public LB to a modest position in private <br> * This region is also sparse, indicating local issues rather than a global one  |\n| R4| * I wish to highlight a sticky pattern here in a region far away from the medal zone <br> * Is this off a public kernel? |\n| R5| * This is similar to R3 but occurs outside the medal positions, necessitating a separate insinuation <br>* This region is more dense than the one in R3 |\n\n<br>All in all, we have a stable leaderboard with few instances of local movements, but not much in terms of churn.\n\n### Analysis of medal positions\n\n| Medal label | Last rank  |Comments| Stability ratio | \n| --- | --- | ------| -----|\n|Gold  |  12 | * Stable teams = 10 <br> * Teams entering from below = 2|  10/12 = 83.33% | \n|Silver |  50| * Stable teams = 31 <br> * Teams entering from below = 6 <br> * Teams entering from above = 1 |   31/38 = 81.58%| \n|Bronze |  100 | * Stable teams = 33 <br> * Teams entering from below =  11<br> * Teams entering from above = 6 | 33/50 = 66.00%  | \n\n<br>This table above reinforces a very small extent of local churn and reiterates the overall LB stability. \n\nI wish to extend sincere wishes to the participants for their onward journey and happy learning to one and all!\n\nRegards,\nRavi Ramakrishnan\n\n",
      "votes": 29
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2854049": "Hello everyone,\n\nI wish to extend sincere congratulations to the winners of the competition and best wishes to all other participants. I hope we could gather some collective insights from the visual below regarding the extent of the churn at the end of the challenge-\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F8273630%2F9e0d1bba56371e9ab69c9ae635d60b5a%2FLB.png?generation=1717479176963787&alt=media)\n\n<br>We have some interesting regions to study further- \n\n| Region Label | Comments   |\n| --- | --- |\n| R1 | * This roughly indicates the medal zone <br> * We observe some extent of local shakeup but it looks stable overall  |\n| R2| * These participants beat the churn and moved into the medal zone <br> * This region is very sparse and indicates the overall LB stability  |\n| R3| * This is the inverse of region R2 <br> * These participated moved down from a medal winning position in the public LB to a modest position in private <br> * This region is also sparse, indicating local issues rather than a global one  |\n| R4| * I wish to highlight a sticky pattern here in a region far away from the medal zone <br> * Is this off a public kernel? |\n| R5| * This is similar to R3 but occurs outside the medal positions, necessitating a separate insinuation <br>* This region is more dense than the one in R3 |\n\n<br>All in all, we have a stable leaderboard with few instances of local movements, but not much in terms of churn.\n\n### Analysis of medal positions\n\n| Medal label | Last rank  |Comments| Stability ratio | \n| --- | --- | ------| -----|\n|Gold  |  12 | * Stable teams = 10 <br> * Teams entering from below = 2|  10/12 = 83.33% | \n|Silver |  50| * Stable teams = 31 <br> * Teams entering from below = 6 <br> * Teams entering from above = 1 |   31/38 = 81.58%| \n|Bronze |  100 | * Stable teams = 33 <br> * Teams entering from below =  11<br> * Teams entering from above = 6 | 33/50 = 66.00%  | \n\n<br>This table above reinforces a very small extent of local churn and reiterates the overall LB stability. \n\nI wish to extend sincere wishes to the participants for their onward journey and happy learning to one and all!\n\nRegards,\nRavi Ramakrishnan\n\n"
  }
}