What is 'songtype_id' exactly? Why it has only values of 1 and 4?
", "messages": [ { "id": 1082461, "postDate": "2020-11-17T23:14:32.933Z", "content": "What is 'songtype_id' exactly? Why it has only values of 1 and 4?
", "rawMarkdown": "What is 'songtype_id' exactly? Why it has only values of 1 and 4? ", "votes": 8 }, { "id": 1082467, "postDate": "2020-11-17T23:25:35.017Z", "content": "Hello, a couple species have training data for two different common song variations. However, the evaluation is only based on predicting species presence (not song type). There is no significance to the songtype_id value (1 vs 4)
", "rawMarkdown": "Hello, a couple species have training data for two different common song variations. However, the evaluation is only based on predicting species presence (not song type). There is no significance to the songtype_id value (1 vs 4)", "votes": 3, "isDeleted": true, "replies": [ { "id": 1082473, "postDate": "2020-11-17T23:40:37.753Z", "content": "@jacklebien Thank you for your answer! The question is what are exactly these common song type variations?
", "rawMarkdown": "@jacklebien Thank you for your answer! The question is what are exactly these common song type variations?" }, { "id": 1082477, "postDate": "2020-11-17T23:55:17.373Z", "content": "An expert biologist on our team determined the most common song type from each species and assigned it songtype_id 1. I understand that songtype_id 4 indicates another (less) common song of the species. I will see if our biologists have more details.
It should be intuitive to understand songtype_id in train_tp.csv: the audio clip in the audio file belongs to a specific song type. HOWEVER, it is hard for me to understand songtype_id in train_fp.csv. What does songtype_id mean in trian_fp.csv? Is the songtype_id correct or the opposite? It doesn't sound intuitive for me since when I do the EDA, I found songtype_id 1 has more, either in tp or fp csv
", "rawMarkdown": "It should be intuitive to understand songtype_id in train_tp.csv: the audio clip in the audio file belongs to a specific song type. HOWEVER, it is hard for me to understand songtype_id in train_fp.csv. What does songtype_id mean in trian_fp.csv? Is the songtype_id correct or the opposite? It doesn't sound intuitive for me since when I do the EDA, I found songtype_id 1 has more, either in tp or fp csv" }, { "id": 1099043, "postDate": "2020-12-02T03:30:35.580Z", "content": "Lets say it is a 'call/muttering/warning/sqabble' etc etc pretty much all living things do…for eg 'I am bored' , 'there is danger looming' etc etc..an expession of inner feelings… As the competition host said , dont think we need to give too much importance to this ..but can definitely keep it as a feature , just in case ..probably to help model learn another distinguishing tie breaking feature..
", "rawMarkdown": "Lets say it is a 'call/muttering/warning/sqabble' etc etc pretty much all living things do...for eg 'I am bored' , 'there is danger looming' etc etc..an expession of inner feelings... As the competition host said , dont think we need to give too much importance to this ..but can definitely keep it as a feature , just in case ..probably to help model learn another distinguishing tie breaking feature.. ", "votes": 1 }, { "id": 1099735, "postDate": "2020-12-02T15:14:52.227Z", "content": "Hello @wubinbai, as mentioned above, songtype_id 1 was assigned to the most common type of call, so that is why most data is songtype_id 1. The songtype_id in train_fp.csv indicates the specific songtype for each species that was confirmed not present in each time-frequency box.
Hello, a couple species have training data for two different common song variations. However, the evaluation is only based on predicting species presence (not song type). There is no significance to the songtype_id value (1 vs 4)
", "votes": 3, "replies": [ { "id": 1082473, "author_name": "Sina Mehdinia", "author_url": "", "post_date": "2020-11-17T23:40:37.753000", "content": "@jacklebien Thank you for your answer! The question is what are exactly these common song type variations?
", "votes": 0, "replies": [] }, { "id": 1082477, "author_name": "", "author_url": "", "post_date": "2020-11-17T23:55:17.373000", "content": "An expert biologist on our team determined the most common song type from each species and assigned it songtype_id 1. I understand that songtype_id 4 indicates another (less) common song of the species. I will see if our biologists have more details.
It should be intuitive to understand songtype_id in train_tp.csv: the audio clip in the audio file belongs to a specific song type. HOWEVER, it is hard for me to understand songtype_id in train_fp.csv. What does songtype_id mean in trian_fp.csv? Is the songtype_id correct or the opposite? It doesn't sound intuitive for me since when I do the EDA, I found songtype_id 1 has more, either in tp or fp csv
", "votes": 0, "replies": [] }, { "id": 1099043, "author_name": "Sunil Krishnan", "author_url": "", "post_date": "2020-12-02T03:30:35.580000", "content": "Lets say it is a 'call/muttering/warning/sqabble' etc etc pretty much all living things do…for eg 'I am bored' , 'there is danger looming' etc etc..an expession of inner feelings… As the competition host said , dont think we need to give too much importance to this ..but can definitely keep it as a feature , just in case ..probably to help model learn another distinguishing tie breaking feature..
", "votes": 1, "replies": [] }, { "id": 1099735, "author_name": "", "author_url": "", "post_date": "2020-12-02T15:14:52.227000", "content": "Hello @wubinbai, as mentioned above, songtype_id 1 was assigned to the most common type of call, so that is why most data is songtype_id 1. The songtype_id in train_fp.csv indicates the specific songtype for each species that was confirmed not present in each time-frequency box.