![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F8273630%2F5643155283a399e30c9a5bf9ef931e49%2FRNA.jpg?generation=1702016827519889&alt=media\")
Hello all,
\nI visualized the leaderboard dynamics and have the below inferences. I split the leaderboard into the below 5 regions highlighting collective behavioral traits-
\n
| Region | \nSub-region | \nInference/ Comments | \n
|---|---|---|
| Region 1 | \n\n | * Relatively stable and top solutions, enabling a medal in the competition. Hearty congratulations to these participants | \n
| Region 2 | \n\n | * This is an interesting region as it perhaps indicates submissions off public work/ similar ideas with little modification. * These submissions elicit stickiness across various ranks and scores | \n
| Region 3 | \n3A | \n* These submissions were highly impacted by the churn | \n
| Region 3 | \n3B | \n* These submissions were moderately impacted by the churn | \n
| Region 3 | \n3C | \n* These submissions were impacted by the churn * These participants missed out on medals too | \n
| Region 4 | \n4A | \n* These submissions were positively impacted by the churn * These participants received medals from positions just out of the medal zone | \n
| Region 4 | \n4B | \n* These submissions were highly positively impacted by the churn * These participants received medals from positions far out of the medal zone * Hearty congratulations to these participants | \n
| Region 5 | \n\n | * These submissions were also highly positively impacted by the churn * These participants did not receive medals but ascended highly from a moderate public LB position to a good private LB position * Hearty congratulations to these participants as well | \n
Happy learning and best wishes!
Hello all,
\nI visualized the leaderboard dynamics and have the below inferences. I split the leaderboard into the below 5 regions highlighting collective behavioral traits-
\n
| Region | \nSub-region | \nInference/ Comments | \n
|---|---|---|
| Region 1 | \n\n | * Relatively stable and top solutions, enabling a medal in the competition. Hearty congratulations to these participants | \n
| Region 2 | \n\n | * This is an interesting region as it perhaps indicates submissions off public work/ similar ideas with little modification. * These submissions elicit stickiness across various ranks and scores | \n
| Region 3 | \n3A | \n* These submissions were highly impacted by the churn | \n
| Region 3 | \n3B | \n* These submissions were moderately impacted by the churn | \n
| Region 3 | \n3C | \n* These submissions were impacted by the churn * These participants missed out on medals too | \n
| Region 4 | \n4A | \n* These submissions were positively impacted by the churn * These participants received medals from positions just out of the medal zone | \n
| Region 4 | \n4B | \n* These submissions were highly positively impacted by the churn * These participants received medals from positions far out of the medal zone * Hearty congratulations to these participants | \n
| Region 5 | \n\n | * These submissions were also highly positively impacted by the churn * These participants did not receive medals but ascended highly from a moderate public LB position to a good private LB position * Hearty congratulations to these participants as well | \n
Happy learning and best wishes!
Dear @ravi20076
\nThe Region 2 look like isoscore strings
\nAll the best,
\ncarl
Thanks for sharing your analysis.
\nI think its a nice game of overfitting and underfitting.
\nThe ranks in region 3A and 4B are very interesting indeed.
My guess- 3A had good models (probably including bpp) with some variant of absolute positional encoding. 4B had not-so-good models, probably also without bpp, but with relative positional encoding or some variant of CNN that could generalize to longer seqs.
", "rawMarkdown": "My guess- 3A had good models (probably including bpp) with some variant of absolute positional encoding. 4B had not-so-good models, probably also without bpp, but with relative positional encoding or some variant of CNN that could generalize to longer seqs.", "votes": null }, { "id": "2553485", "postDate": "12/08/2023 09:39:01", "content": "Thanks for sharing @ravi20076. Very nice analysis ✨
", "rawMarkdown": "Thanks for sharing @ravi20076. Very nice analysis ✨", "votes": null }, { "id": "2553621", "postDate": "12/08/2023 11:58:43", "content": "Thanks for sharing @ravi20076, this is interesting!
", "rawMarkdown": "Thanks for sharing @ravi20076, this is interesting!", "votes": null }, { "id": "2554107", "postDate": "12/08/2023 19:35:21", "content": "Thanks for the analysis.
", "rawMarkdown": "Thanks for the analysis.", "votes": null }, { "id": "2554491", "postDate": "12/09/2023 06:54:34", "content": "Nice analysis sir! Very insightful
", "rawMarkdown": "Nice analysis sir! Very insightful", "votes": null }, { "id": "2555425", "postDate": "12/09/2023 21:58:40", "content": "Specifically, I think it corresponds to the notebooks with scores:
\nPublic 0.15472 Private 0.26885
\nPublic 0.18588 Private 0.23524
\nPublic 0.33975 Private 0.33290
Beautiful Visualization! Keep it up Ravi! Was this done in R?
", "rawMarkdown": "Beautiful Visualization! Keep it up Ravi! Was this done in R?", "votes": null }, { "id": "2556106", "postDate": "12/10/2023 13:19:14", "content": "Great analysis @ravi20076
", "rawMarkdown": "Great analysis @ravi20076", "votes": null }, { "id": "2556934", "postDate": "12/11/2023 05:01:06", "content": "No it was done in excel @dtrade84
", "rawMarkdown": "No it was done in excel @dtrade84", "votes": null }, { "id": "2560407", "postDate": "12/13/2023 15:41:36", "content": "Great analysis! Can you please define the meaning of churn in this context?
", "rawMarkdown": "Great analysis! Can you please define the meaning of churn in this context?", "votes": null }, { "id": "2563777", "postDate": "12/16/2023 15:21:45", "content": "@seanbearden this is just a change from public-private scoring.
", "rawMarkdown": "seanbearden this is just a change from public-private scoring.", "votes": null } ], "comments": [ { "id": 2553328, "author_name": "carlmcbrideellis", "author_url": "", "post_date": "12/08/2023 06:44:54", "content": "Dear @ravi20076
\nThe Region 2 look like isoscore strings
\nAll the best,
\ncarl
Specifically, I think it corresponds to the notebooks with scores:
\nPublic 0.15472 Private 0.26885
\nPublic 0.18588 Private 0.23524
\nPublic 0.33975 Private 0.33290
Thanks for sharing your analysis.
\nI think its a nice game of overfitting and underfitting.
\nThe ranks in region 3A and 4B are very interesting indeed.
My guess- 3A had good models (probably including bpp) with some variant of absolute positional encoding. 4B had not-so-good models, probably also without bpp, but with relative positional encoding or some variant of CNN that could generalize to longer seqs.
", "votes": null, "replies": [] } ] }, { "id": 2553485, "author_name": "redpen12", "author_url": "", "post_date": "12/08/2023 09:39:01", "content": "Thanks for sharing @ravi20076. Very nice analysis ✨
", "votes": null, "replies": [] }, { "id": 2553621, "author_name": "pranshubahadur", "author_url": "", "post_date": "12/08/2023 11:58:43", "content": "Thanks for sharing @ravi20076, this is interesting!
", "votes": null, "replies": [] }, { "id": 2554107, "author_name": "nikhil9084", "author_url": "", "post_date": "12/08/2023 19:35:21", "content": "Thanks for the analysis.
", "votes": null, "replies": [] }, { "id": 2554491, "author_name": "abhasmalguri", "author_url": "", "post_date": "12/09/2023 06:54:34", "content": "Nice analysis sir! Very insightful
", "votes": null, "replies": [] }, { "id": 2555506, "author_name": "dtrade84", "author_url": "", "post_date": "12/10/2023 01:15:28", "content": "Beautiful Visualization! Keep it up Ravi! Was this done in R?
", "votes": null, "replies": [ { "id": 2556934, "author_name": "ravi20076", "author_url": "", "post_date": "12/11/2023 05:01:06", "content": "No it was done in excel @dtrade84
", "votes": null, "replies": [] } ] }, { "id": 2556106, "author_name": "dishaagarwal7001", "author_url": "", "post_date": "12/10/2023 13:19:14", "content": "Great analysis @ravi20076
", "votes": null, "replies": [] }, { "id": 2560407, "author_name": "seanbearden", "author_url": "", "post_date": "12/13/2023 15:41:36", "content": "Great analysis! Can you please define the meaning of churn in this context?
", "votes": null, "replies": [ { "id": 2563777, "author_name": "ravi20076", "author_url": "", "post_date": "12/16/2023 15:21:45", "content": "@seanbearden this is just a change from public-private scoring.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2553322": "Hello all,\n\nI visualized the leaderboard dynamics and have the below inferences. I split the leaderboard into the below 5 regions highlighting collective behavioral traits-\n\n\n\n| Region | Sub-region| Inference/ Comments |\n| --- | --- |---|\n| Region 1 ||* Relatively stable and top solutions, enabling a medal in the competition. Hearty congratulations to these participants |\n| Region 2 | |* This is an interesting region as it perhaps indicates submissions off public work/ similar ideas with little modification.