I would like to discuss ideas about features not directly available on the competition dataset. Some of the ideas I have are:
\nI would like to discuss ideas about features not directly available on the competition dataset. Some of the ideas I have are:
\nFacebook + Instragram followers and more engagement features as replies, likes ….
", "rawMarkdown": "Facebook + Instragram followers and more engagement features as replies, likes ....", "votes": null }, { "id": "1383402", "postDate": "07/10/2021 19:48:27", "content": "Definitely been thinking of this as they're obvious basic digital engagement metrics. Bit concerned though that we cannot use live versions of these on the private test so might be useless. Have you tried using these in your models @enric1296 ? :)
", "rawMarkdown": "Definitely been thinking of this as they're obvious basic digital engagement metrics. Bit concerned though that we cannot use live versions of these on the private test so might be useless. Have you tried using these in your models @enric1296 ? :)", "votes": null }, { "id": "1383849", "postDate": "07/11/2021 09:25:50", "content": "Not yet, but if you get that data would be enough by having the player + Date aggregation since 2018 until 2021-05. even if its montly aggregated woould be effective
", "rawMarkdown": "Not yet, but if you get that data would be enough by having the player + Date aggregation since 2018 until 2021-05. even if its montly aggregated woould be effective", "votes": null }, { "id": "1385762", "postDate": "07/13/2021 02:53:59", "content": "Do you know where that can be scraped/downloaded?
", "rawMarkdown": "Do you know where that can be scraped/downloaded?", "votes": null }, { "id": "1478503", "postDate": "08/18/2021 03:14:00", "content": "@jacobhowardparker Did you end up doing that to achieve your high performance? Congrats so far btw!
", "rawMarkdown": "jacobhowardparker Did you end up doing that to achieve your high performance? Congrats so far btw!", "votes": null }, { "id": "1481754", "postDate": "08/19/2021 16:50:48", "content": "No I didn't use these. And thankyou! Let's see what happens over the next month haha
", "rawMarkdown": "No I didn't use these. And thankyou! Let's see what happens over the next month haha", "votes": null }, { "id": "1481808", "postDate": "08/19/2021 17:16:18", "content": "These are some good ideas. As @jacobhowardparker mentioned, the block on live updates for the test period limits their effectiveness. They could serve as foundational measures though, similar to playing for a popular team. Here's what I tried:
\nOnly ADP showed a gain, and that was at first. It didn't have a noticeable effect after we added more features.
", "rawMarkdown": "These are some good ideas. As @jacobhowardparker mentioned, the block on live updates for the test period limits their effectiveness. They could serve as foundational measures though, similar to playing for a popular team. Here's what I tried:\n- TV market size\n- Team TV revenue\n- Average game attendance\n- Team payroll spend\n- Fantasy football stats including Average Draft Position (ADP)\n\nOnly ADP showed a gain, and that was at first. It didn't have a noticeable effect after we added more features.", "votes": null } ], "comments": [ { "id": 1382845, "author_name": "enric1296", "author_url": "", "post_date": "07/10/2021 09:05:15", "content": "Facebook + Instragram followers and more engagement features as replies, likes ….
", "votes": null, "replies": [ { "id": 1383402, "author_name": "jacobhowardparker", "author_url": "", "post_date": "07/10/2021 19:48:27", "content": "Definitely been thinking of this as they're obvious basic digital engagement metrics. Bit concerned though that we cannot use live versions of these on the private test so might be useless. Have you tried using these in your models @enric1296 ? :)
", "votes": null, "replies": [] }, { "id": 1383849, "author_name": "enric1296", "author_url": "", "post_date": "07/11/2021 09:25:50", "content": "Not yet, but if you get that data would be enough by having the player + Date aggregation since 2018 until 2021-05. even if its montly aggregated woould be effective
", "votes": null, "replies": [] }, { "id": 1385762, "author_name": "kaito510", "author_url": "", "post_date": "07/13/2021 02:53:59", "content": "Do you know where that can be scraped/downloaded?
", "votes": null, "replies": [] }, { "id": 1478503, "author_name": "kaito510", "author_url": "", "post_date": "08/18/2021 03:14:00", "content": "@jacobhowardparker Did you end up doing that to achieve your high performance? Congrats so far btw!
", "votes": null, "replies": [] }, { "id": 1481754, "author_name": "jacobhowardparker", "author_url": "", "post_date": "08/19/2021 16:50:48", "content": "No I didn't use these. And thankyou! Let's see what happens over the next month haha
", "votes": null, "replies": [] } ] }, { "id": 1481808, "author_name": "jpmiller", "author_url": "", "post_date": "08/19/2021 17:16:18", "content": "These are some good ideas. As @jacobhowardparker mentioned, the block on live updates for the test period limits their effectiveness. They could serve as foundational measures though, similar to playing for a popular team. Here's what I tried:
\nOnly ADP showed a gain, and that was at first. It didn't have a noticeable effect after we added more features.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1382765": "I would like to discuss ideas about features not directly available on the competition dataset. Some of the ideas I have are:\n\n1. Day of week and time of clash. Fan engagement should change depending on if the game is played on a weekday or during the weekend. Important to consider the different time zones for the different cities the teams represent.\n2. Fantasy baseball statistics. Fans that play the fantasy competitions on mlb.com (e.g. Beat the Streak, Fantasy Baseball etc) will be engaged to players that are popular for these competitions.\n3. Rivalries. Everyone knows fan engagement for NYY vs BOS will be high compared to other BOS matches.\n4. Player salaries. Everyone knows in this day and age, this is important. I have provided this data here: https://www.kaggle.com/kaito510/mlb-player-salaries", "1382845": "Facebook + Instragram followers and more engagement features as replies, likes ....", "1383402": "Definitely been thinking of this as they're obvious basic digital engagement metrics. Bit concerned though that we cannot use live versions of these on the private test so might be useless. Have you tried using these in your models @enric1296 ? :)", "1383849": "Not yet, but if you get that data would be enough by having the player + Date aggregation since 2018 until 2021-05. even if its montly aggregated woould be effective", "1385762": "Do you know where that can be scraped/downloaded?", "1478503": "jacobhowardparker Did you end up doing that to achieve your high performance? Congrats so far btw!", "1481754": "No I didn't use these. And thankyou! Let's see what happens over the next month haha", "1481808": "These are some good ideas. As @jacobhowardparker mentioned, the block on live updates for the test period limits their effectiveness. They could serve as foundational measures though, similar to playing for a popular team. Here's what I tried:\n- TV market size\n- Team TV revenue\n- Average game attendance\n- Team payroll spend\n- Fantasy football stats including Average Draft Position (ADP)\n\nOnly ADP showed a gain, and that was at first. It didn't have a noticeable effect after we added more features." }, "source": "meta" }