One of the biggest unknown details of the dataset is the instance each horse crossed the finish line and thus finished the race.
\nThis is because the horses' locations are still being recorded even after they have crossed the finish line, up until the last horse finishes. It is evident by the fact that there is an equal number of location samples for every horse that finished the race.
\nThis poses a problem since the winning positions are not determined by the horses positions at the last location samples, but rather on how soon they had crossed the finish line.
\nConsequently, relative horses positions at the last trakus indices often do not match the race finishing places.
\nFurthermore, calculations such as average speed during the race should ignore the part of the race after the horses crossed the finish line.
To solve this problem, I have developed a method to determine the trakus index in which each participating horse finished the race.
\nIt is based on the assumption that the location recording stops at the moment the last horse finishes the race, hence the finish line is the last horse's location in the final trakus index.
\nKnowing this location, the instance each horse finishes the race is determined by comparing its locations to the calculated finish line, and finding when the horse first surpassed it.
There are a few things to note about this method:
\nI created the following notebook with my calculation:
\n[https://www.kaggle.com/shapital/calculating-when-horses-finish-the-race]
Feel free to use it, ask questions and propose improvements.
", "messages": [ { "id": "1968912", "postDate": "10/03/2022 09:11:51", "content": "One of the biggest unknown details of the dataset is the instance each horse crossed the finish line and thus finished the race.
\nThis is because the horses' locations are still being recorded even after they have crossed the finish line, up until the last horse finishes. It is evident by the fact that there is an equal number of location samples for every horse that finished the race.
\nThis poses a problem since the winning positions are not determined by the horses positions at the last location samples, but rather on how soon they had crossed the finish line.
\nConsequently, relative horses positions at the last trakus indices often do not match the race finishing places.
\nFurthermore, calculations such as average speed during the race should ignore the part of the race after the horses crossed the finish line.
To solve this problem, I have developed a method to determine the trakus index in which each participating horse finished the race.
\nIt is based on the assumption that the location recording stops at the moment the last horse finishes the race, hence the finish line is the last horse's location in the final trakus index.
\nKnowing this location, the instance each horse finishes the race is determined by comparing its locations to the calculated finish line, and finding when the horse first surpassed it.
There are a few things to note about this method:
\nI created the following notebook with my calculation:
\n[https://www.kaggle.com/shapital/calculating-when-horses-finish-the-race]
Feel free to use it, ask questions and propose improvements.
", "rawMarkdown": "One of the biggest unknown details of the dataset is the instance each horse crossed the finish line and thus finished the race.\nThis is because the horses' locations are still being recorded even after they have crossed the finish line, up until the last horse finishes. It is evident by the fact that there is an equal number of location samples for every horse that finished the race.\nThis poses a problem since the winning positions are not determined by the horses positions at the last location samples, but rather on how soon they had crossed the finish line.\nConsequently, relative horses positions at the last trakus indices often do not match the race finishing places.\nFurthermore, calculations such as average speed during the race should ignore the part of the race after the horses crossed the finish line.\n\nTo solve this problem, I have developed a method to determine the trakus index in which each participating horse finished the race.\nIt is based on the assumption that the location recording stops at the moment the last horse finishes the race, hence the finish line is the last horse's location in the final trakus index.\nKnowing this location, the instance each horse finishes the race is determined by comparing its locations to the calculated finish line, and finding when the horse first surpassed it.\n\nThere are a few things to note about this method:\n1. Horses that do not finish the race are not used for calculating the finish line.\n2. Long races contain several laps of the track, therefore the horses cross the finish line more than once.\n3. The low temporal resolution (0.25s) may cause discrepancies between the calculated finishing instances and the actual race results.\n4. The assumption about the location of the finish line may not be accurate.\n\nI created the following notebook with my calculation:\n[https://www.kaggle.com/shapital/calculating-when-horses-finish-the-race]\n\nFeel free to use it, ask questions and propose improvements.", "votes": null }, { "id": "1969798", "postDate": "10/03/2022 17:29:16", "content": "Sir plz tell me which algo to use to autmate it.
", "rawMarkdown": "Sir plz tell me which algo to use to autmate it.", "votes": null }, { "id": "1985937", "postDate": "10/13/2022 17:30:15", "content": "The finish line is the same for all races, it's the starting line that changes based on the distance of the race. There seems to be too much variability in the final locations of the last place horses for that to be the finish line. I tried looking at 4-horse races for the outer track at Aquaduct to reduce the variability. I'm curious about the actual mechanics of how the trakus_index is stopped and started, but if it is a human pushing a button to stop a timer that bound to introduce a lot of variability.
\nBelow is a notebook with my graphs. It's my first notebook ever, so it took me awhile to figure out how to create and post it. It is a R notebook which seems to not be very common in this competition.
\nhttps://www.kaggle.com/code/swhardy/aquaduct-finish-line
", "rawMarkdown": "The finish line is the same for all races, it's the starting line that changes based on the distance of the race. There seems to be too much variability in the final locations of the last place horses for that to be the finish line. I tried looking at 4-horse races for the outer track at Aquaduct to reduce the variability. I'm curious about the actual mechanics of how the trakus_index is stopped and started, but if it is a human pushing a button to stop a timer that bound to introduce a lot of variability.\n\nBelow is a notebook with my graphs. It's my first notebook ever, so it took me awhile to figure out how to create and post it. It is a R notebook which seems to not be very common in this competition.\n\nhttps://www.kaggle.com/code/swhardy/aquaduct-finish-line", "votes": null }, { "id": "2023167", "postDate": "11/09/2022 15:27:44", "content": "The trakus device stops recording 20 trakus indexes (5 seconds) after the last horse reaches the finish line.
\nIf you identify the trakus index, when the last horse crosses the finish line, it appears, that records of the race ends 20 trakus indexes later.
\nIn our model of the finish line (see our notebook Big Data Derby 2022: Statistical model, https://www.kaggle.com/code/adminpretaktovaniesk/big-data-derby-2022-statistical-model) we see that records end 20 trakus indexes after the last horse crosses the finish line.
", "rawMarkdown": "The trakus device stops recording 20 trakus indexes (5 seconds) after the last horse reaches the finish line. \n\nIf you identify the trakus index, when the last horse crosses the finish line, it appears, that records of the race ends 20 trakus indexes later. \n\nIn our model of the finish line (see our notebook Big Data Derby 2022: Statistical model, https://www.kaggle.com/code/adminpretaktovaniesk/big-data-derby-2022-statistical-model) we see that records end 20 trakus indexes after the last horse crosses the finish line.", "votes": null } ], "comments": [ { "id": 1969798, "author_name": "sairamsatish", "author_url": "", "post_date": "10/03/2022 17:29:16", "content": "Sir plz tell me which algo to use to autmate it.
", "votes": null, "replies": [] }, { "id": 1985937, "author_name": "swhardy", "author_url": "", "post_date": "10/13/2022 17:30:15", "content": "The finish line is the same for all races, it's the starting line that changes based on the distance of the race. There seems to be too much variability in the final locations of the last place horses for that to be the finish line. I tried looking at 4-horse races for the outer track at Aquaduct to reduce the variability. I'm curious about the actual mechanics of how the trakus_index is stopped and started, but if it is a human pushing a button to stop a timer that bound to introduce a lot of variability.
\nBelow is a notebook with my graphs. It's my first notebook ever, so it took me awhile to figure out how to create and post it. It is a R notebook which seems to not be very common in this competition.
\nhttps://www.kaggle.com/code/swhardy/aquaduct-finish-line
", "votes": null, "replies": [] }, { "id": 2023167, "author_name": "adminpretaktovaniesk", "author_url": "", "post_date": "11/09/2022 15:27:44", "content": "The trakus device stops recording 20 trakus indexes (5 seconds) after the last horse reaches the finish line.
\nIf you identify the trakus index, when the last horse crosses the finish line, it appears, that records of the race ends 20 trakus indexes later.
\nIn our model of the finish line (see our notebook Big Data Derby 2022: Statistical model, https://www.kaggle.com/code/adminpretaktovaniesk/big-data-derby-2022-statistical-model) we see that records end 20 trakus indexes after the last horse crosses the finish line.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1968912": "One of the biggest unknown details of the dataset is the instance each horse crossed the finish line and thus finished the race.\nThis is because the horses' locations are still being recorded even after they have crossed the finish line, up until the last horse finishes. It is evident by the fact that there is an equal number of location samples for every horse that finished the race.\nThis poses a problem since the winning positions are not determined by the horses positions at the last location samples, but rather on how soon they had crossed the finish line.\nConsequently, relative horses positions at the last trakus indices often do not match the race finishing places.\nFurthermore, calculations such as average speed during the race should ignore the part of the race after the horses crossed the finish line.\n\nTo solve this problem, I have developed a method to determine the trakus index in which each participating horse finished the race.\nIt is based on the assumption that the location recording stops at the moment the last horse finishes the race, hence the finish line is the last horse's location in the final trakus index.\nKnowing this location, the instance each horse finishes the race is determined by comparing its locations to the calculated finish line, and finding when the horse first surpassed it.\n\nThere are a few things to note about this method:\n1. Horses that do not finish the race are not used for calculating the finish line.\n2. Long races contain several laps of the track, therefore the horses cross the finish line more than once.\n3. The low temporal resolution (0.25s) may cause discrepancies between the calculated finishing instances and the actual race results.\n4. The assumption about the location of the finish line may not be accurate.\n\nI created the following notebook with my calculation:\n[https://www.kaggle.com/shapital/calculating-when-horses-finish-the-race]\n\nFeel free to use it, ask questions and propose improvements.", "1969798": "Sir plz tell me which algo to use to autmate it.", "1985937": "The finish line is the same for all races, it's the starting line that changes based on the distance of the race. There seems to be too much variability in the final locations of the last place horses for that to be the finish line. I tried looking at 4-horse races for the outer track at Aquaduct to reduce the variability. I'm curious about the actual mechanics of how the trakus_index is stopped and started, but if it is a human pushing a button to stop a timer that bound to introduce a lot of variability.\n\nBelow is a notebook with my graphs. It's my first notebook ever, so it took me awhile to figure out how to create and post it. It is a R notebook which seems to not be very common in this competition.\n\nhttps://www.kaggle.com/code/swhardy/aquaduct-finish-line", "2023167": "The trakus device stops recording 20 trakus indexes (5 seconds) after the last horse reaches the finish line. \n\nIf you identify the trakus index, when the last horse crosses the finish line, it appears, that records of the race ends 20 trakus indexes later. \n\nIn our model of the finish line (see our notebook Big Data Derby 2022: Statistical model, https://www.kaggle.com/code/adminpretaktovaniesk/big-data-derby-2022-statistical-model) we see that records end 20 trakus indexes after the last horse crosses the finish line." }, "source": "meta" }