![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F331718%2Fbf018f76aa097021417a318feb288c29%2FScreenshot%20from%202020-02-17%2000-50-46.png?generation=1581929484802795&alt=media\")
I'm looking through the event data in the MEvents2015 through MEvents2019 files. I noticed that there are some rows with EventTeamID=0. I attached a csv with the subset of events with this condition. Here's a screenshot of what that data looks like.
The data page says
\n\n\n\n\nEventTeamID - this is the ID of the team that the event is logged for, which will either be the WTeamID or the LTeamID.
\n
But when EventTeamID=0 neither WTeamID=0 nor LTeamID=0. What does EventTeamID=0 mean?
I'm looking through the event data in the MEvents2015 through MEvents2019 files. I noticed that there are some rows with EventTeamID=0. I attached a csv with the subset of events with this condition. Here's a screenshot of what that data looks like.
The data page says
\n\n\n\n\nEventTeamID - this is the ID of the team that the event is logged for, which will either be the WTeamID or the LTeamID.
\n
But when EventTeamID=0 neither WTeamID=0 nor LTeamID=0. What does EventTeamID=0 mean?
For the 2019 data in your screenshot, that looks correct because it is for commercial timeouts or starting a jump ball at the beginning of a game. Neither of those is associated with a team; they are overall game events, included in there in order to help illustrate the game status. For the 2015 data, that indeed looks wrong and is likely from bad original data. I will have to look into it.
", "rawMarkdown": "For the 2019 data in your screenshot, that looks correct because it is for commercial timeouts or starting a jump ball at the beginning of a game. Neither of those is associated with a team; they are overall game events, included in there in order to help illustrate the game status. For the 2015 data, that indeed looks wrong and is likely from bad original data. I will have to look into it.", "votes": null }, { "id": "748639", "postDate": "02/17/2020 19:20:25", "content": "Thanks @jeffsonas for the quick response. The overall game events make sense.
", "rawMarkdown": "Thanks @jeffsonas for the quick response. The overall game events make sense.", "votes": null }, { "id": "748661", "postDate": "02/17/2020 19:50:20", "content": "It's only for season=2019+ that those overall game event records will be in there. From earlier seasons, it seems like the bulk of the Team=0 data is for Morgan St games (team ID 1288) from 2015, and Georgia Tech (team ID 1210) from 2016, and a few other games. If it were only giving you the zero for half of the events in such games, then you could obviously try assuming the zero was consistently for one of the teams. But since it seems to be giving zeroes for all rows in those games, then we can't distinguish which team is which in the events, and so you will probably have to just exclude those entire games from your analysis. It may just be bad data originally, or it may be that I can try to preserve the references by changing how I am extracting the data. There's a lot of transformation steps going on. We will probably be doing another pull of the data in a couple weeks, so that we can get 2020 season data out there for people to look at, and so I will try to see if I can recover these games better at that point.
", "rawMarkdown": "It's only for season=2019+ that those overall game event records will be in there. From earlier seasons, it seems like the bulk of the Team=0 data is for Morgan St games (team ID 1288) from 2015, and Georgia Tech (team ID 1210) from 2016, and a few other games. If it were only giving you the zero for half of the events in such games, then you could obviously try assuming the zero was consistently for one of the teams. But since it seems to be giving zeroes for all rows in those games, then we can't distinguish which team is which in the events, and so you will probably have to just exclude those entire games from your analysis. It may just be bad data originally, or it may be that I can try to preserve the references by changing how I am extracting the data. There's a lot of transformation steps going on. We will probably be doing another pull of the data in a couple weeks, so that we can get 2020 season data out there for people to look at, and so I will try to see if I can recover these games better at that point.", "votes": null }, { "id": "748687", "postDate": "02/17/2020 20:52:40", "content": "Got it. Sounds good to me.
", "rawMarkdown": "Got it. Sounds good to me.", "votes": null } ], "comments": [ { "id": 748608, "author_name": "jeffsonas", "author_url": "", "post_date": "02/17/2020 18:29:11", "content": "For the 2019 data in your screenshot, that looks correct because it is for commercial timeouts or starting a jump ball at the beginning of a game. Neither of those is associated with a team; they are overall game events, included in there in order to help illustrate the game status. For the 2015 data, that indeed looks wrong and is likely from bad original data. I will have to look into it.
", "votes": null, "replies": [ { "id": 748639, "author_name": "mmotoki", "author_url": "", "post_date": "02/17/2020 19:20:25", "content": "Thanks @jeffsonas for the quick response. The overall game events make sense.
", "votes": null, "replies": [] }, { "id": 748661, "author_name": "jeffsonas", "author_url": "", "post_date": "02/17/2020 19:50:20", "content": "It's only for season=2019+ that those overall game event records will be in there. From earlier seasons, it seems like the bulk of the Team=0 data is for Morgan St games (team ID 1288) from 2015, and Georgia Tech (team ID 1210) from 2016, and a few other games. If it were only giving you the zero for half of the events in such games, then you could obviously try assuming the zero was consistently for one of the teams. But since it seems to be giving zeroes for all rows in those games, then we can't distinguish which team is which in the events, and so you will probably have to just exclude those entire games from your analysis. It may just be bad data originally, or it may be that I can try to preserve the references by changing how I am extracting the data. There's a lot of transformation steps going on. We will probably be doing another pull of the data in a couple weeks, so that we can get 2020 season data out there for people to look at, and so I will try to see if I can recover these games better at that point.
", "votes": null, "replies": [] }, { "id": 748687, "author_name": "mmotoki", "author_url": "", "post_date": "02/17/2020 20:52:40", "content": "Got it. Sounds good to me.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "748186": "I'm looking through the event data in the `MEvents2015` through `MEvents2019` files. I noticed that there are some rows with `EventTeamID=0`. I attached a csv with the subset of events with this condition. Here's a screenshot of what that data looks like. \n\n\n\nThe data page says\n> EventTeamID - this is the ID of the team that the event is logged for, which will either be the WTeamID or the LTeamID.\n\nBut when `EventTeamID=0` neither `WTeamID=0` nor `LTeamID=0`. What does `EventTeamID=0` mean?", "748608": "For the 2019 data in your screenshot, that looks correct because it is for commercial timeouts or starting a jump ball at the beginning of a game. Neither of those is associated with a team; they are overall game events, included in there in order to help illustrate the game status. For the 2015 data, that indeed looks wrong and is likely from bad original data. I will have to look into it.", "748639": "Thanks @jeffsonas for the quick response. The overall game events make sense.", "748661": "It's only for season=2019+ that those overall game event records will be in there. From earlier seasons, it seems like the bulk of the Team=0 data is for Morgan St games (team ID 1288) from 2015, and Georgia Tech (team ID 1210) from 2016, and a few other games. If it were only giving you the zero for half of the events in such games, then you could obviously try assuming the zero was consistently for one of the teams. But since it seems to be giving zeroes for all rows in those games, then we can't distinguish which team is which in the events, and so you will probably have to just exclude those entire games from your analysis. It may just be bad data originally, or it may be that I can try to preserve the references by changing how I am extracting the data. There's a lot of transformation steps going on. We will probably be doing another pull of the data in a couple weeks, so that we can get 2020 season data out there for people to look at, and so I will try to see if I can recover these games better at that point.", "748687": "Got it. Sounds good to me." }, "source": "meta" }