![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2875376%2Fdfb581cf866c9006531f7b37c13044a4%2FScreen%20Shot%202024-10-23%20at%206.58.06%20PM.png?generation=1729690122179305&alt=media\")
Hey, there! I did network analysis of tags using Gephi and NetworkX.
\nHere's the notebook: https://www.kaggle.com/code/ahsuna123/tag-network-analysis-networkx-gephi
\nI hope the notebook helps in getting an idea of what features share similarity.
I came across these 4 clusters with tags 3, 7, 14 and 12:
![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2875376%2F5af6f3f7fa2d304b6742c080f3c8811a%2FScreen%20Shot%202024-10-23%20at%206.57.13%20PM.png?generation=1729690075682317&alt=media\")
Hey, there! I did network analysis of tags using Gephi and NetworkX.
\nHere's the notebook: https://www.kaggle.com/code/ahsuna123/tag-network-analysis-networkx-gephi
\nI hope the notebook helps in getting an idea of what features share similarity.
I came across these 4 clusters with tags 3, 7, 14 and 12:
@ahsuna123 Looks great ! Thank you for sharing !
", "rawMarkdown": "ahsuna123 Looks great ! Thank you for sharing !", "votes": null }, { "id": "3027282", "postDate": "10/24/2024 16:21:42", "content": "Clustering key features using Gephi and NetworkX revealed strong correlations in tags 3, 7, 14, and 12, offering deeper insights for improving market data forecasting accuracy.
", "rawMarkdown": "Clustering key features using Gephi and NetworkX revealed strong correlations in tags 3, 7, 14, and 12, offering deeper insights for improving market data forecasting accuracy.", "votes": null }, { "id": "3028831", "postDate": "10/26/2024 14:56:28", "content": "I just came across your great work.
\nI am just wondering how you can use this result to improve your model?
Hey! @farhankardan
\nGreat Question! This might not be used directly to improve the results but it would help in De-Anonymization of the features. As mentioned at the end of the notebook as well:
\n", "rawMarkdown": "Hey! @farhankardan \nGreat Question! This might not be used directly to improve the results but it would help in De-Anonymization of the features. As mentioned at the end of the notebook as well:\n> Running each of these sets of feature groups through a non-linear dimension reduction algorithm could help derive a cleaner signal. These feature groups represent properties of the underlying data probably scaled over different time and/or modal domains.\nBy isolating these feature groups and examining how they relate to each other, we can engineer better features.", "votes": null }, { "id": "3029628", "postDate": "10/27/2024 14:20:10", "content": "Running each of these sets of feature groups through a non-linear dimension reduction algorithm could help derive a cleaner signal. These feature groups represent properties of the underlying data probably scaled over different time and/or modal domains.
\n
\n By isolating these feature groups and examining how they relate to each other, we can engineer better features.
Great stuff! Working on a similar approach!
", "rawMarkdown": "Great stuff! Working on a similar approach!", "votes": null }, { "id": "3029754", "postDate": "10/27/2024 17:35:25", "content": "Glad to hear that, Diganta! :)
", "rawMarkdown": "Glad to hear that, Diganta! :)", "votes": null }, { "id": "3088678", "postDate": "01/05/2025 02:38:52", "content": "Have you done this analysis by symbol?
", "rawMarkdown": "Have you done this analysis by symbol?", "votes": null } ], "comments": [ { "id": 3026151, "author_name": "chumajin", "author_url": "", "post_date": "10/23/2024 13:42:46", "content": "@ahsuna123 Looks great ! Thank you for sharing !
", "votes": null, "replies": [] }, { "id": 3027282, "author_name": "prashantkumaryt", "author_url": "", "post_date": "10/24/2024 16:21:42", "content": "Clustering key features using Gephi and NetworkX revealed strong correlations in tags 3, 7, 14, and 12, offering deeper insights for improving market data forecasting accuracy.
", "votes": null, "replies": [] }, { "id": 3028831, "author_name": "farhankardan", "author_url": "", "post_date": "10/26/2024 14:56:28", "content": "I just came across your great work.
\nI am just wondering how you can use this result to improve your model?
Hey! @farhankardan
\nGreat Question! This might not be used directly to improve the results but it would help in De-Anonymization of the features. As mentioned at the end of the notebook as well:
\n", "votes": null, "replies": [] } ] }, { "id": 3029628, "author_name": "digantabhattacharya", "author_url": "", "post_date": "10/27/2024 14:20:10", "content": "Running each of these sets of feature groups through a non-linear dimension reduction algorithm could help derive a cleaner signal. These feature groups represent properties of the underlying data probably scaled over different time and/or modal domains.
\n
\n By isolating these feature groups and examining how they relate to each other, we can engineer better features.
Great stuff! Working on a similar approach!
", "votes": null, "replies": [ { "id": 3029754, "author_name": "ahsuna123", "author_url": "", "post_date": "10/27/2024 17:35:25", "content": "Glad to hear that, Diganta! :)
", "votes": null, "replies": [] } ] }, { "id": 3088678, "author_name": "patriciogalvan", "author_url": "", "post_date": "01/05/2025 02:38:52", "content": "Have you done this analysis by symbol?
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3026137": "Hey, there! I did network analysis of tags using Gephi and NetworkX. \nHere's the notebook: https://www.kaggle.com/code/ahsuna123/tag-network-analysis-networkx-gephi\nI hope the notebook helps in getting an idea of what features share similarity.\n\nI came across these 4 clusters with tags 3, 7, 14 and 12:\n\n", "3026151": "ahsuna123 Looks great ! Thank you for sharing !", "3027282": "Clustering key features using Gephi and NetworkX revealed strong correlations in tags 3, 7, 14, and 12, offering deeper insights for improving market data forecasting accuracy.", "3028831": "I just came across your great work.\nI am just wondering how you can use this result to improve your model?", "3029581": "Hey! @farhankardan \nGreat Question! This might not be used directly to improve the results but it would help in De-Anonymization of the features. As mentioned at the end of the notebook as well:\n> Running each of these sets of feature groups through a non-linear dimension reduction algorithm could help derive a cleaner signal. These feature groups represent properties of the underlying data probably scaled over different time and/or modal domains.\nBy isolating these feature groups and examining how they relate to each other, we can engineer better features.", "3029628": "Great stuff! Working on a similar approach!", "3029754": "Glad to hear that, Diganta! :)", "3088678": "Have you done this analysis by symbol?" }, "source": "meta" }