{
  "id": 239148,
  "title": "Some points worth considering?",
  "url": "/competitions/birdclef-2021/discussion/239148",
  "author_name": "",
  "post_date": "2021-05-15T01:45:00.706894700Z",
  "votes": 4,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I would like to know … if someone wishes to comment on how are you addressing the following points … and if in any way anyone believes that it is worth taking them into account….</p>\n<ul>\n<li>test_soundscapes <strong>approximately</strong> 80 recordings  <strong>roughly</strong> 10 minutes long </li>\n<li>file names include the <strong>date</strong> the recording was taken</li>\n<li>text files with the <strong>name</strong> and <strong>approximate coordinates</strong> of the recording location</li>\n<li><strong>date</strong>s the test set soundscapes were recorded</li>\n</ul>",
  "messages": [
    {
      "id": "1308093",
      "postDate": "05/15/2021 01:45:00",
      "content": "<p>I would like to know … if someone wishes to comment on how are you addressing the following points … and if in any way anyone believes that it is worth taking them into account….</p>\n<ul>\n<li>test_soundscapes <strong>approximately</strong> 80 recordings  <strong>roughly</strong> 10 minutes long </li>\n<li>file names include the <strong>date</strong> the recording was taken</li>\n<li>text files with the <strong>name</strong> and <strong>approximate coordinates</strong> of the recording location</li>\n<li><strong>date</strong>s the test set soundscapes were recorded</li>\n</ul>",
      "rawMarkdown": "I would like to know ... if someone wishes to comment on how are you addressing the following points ... and if in any way anyone believes that it is worth taking them into account....\n\n- test_soundscapes **approximately** 80 recordings  **roughly** 10 minutes long \n- file names include the **date** the recording was taken\n- text files with the **name** and **approximate coordinates** of the recording location\n- **date**s the test set soundscapes were recorded",
      "votes": null
    },
    {
      "id": "1309008",
      "postDate": "05/15/2021 15:33:58",
      "content": "<p>From a birder’s perspective, taking location and date into account is absolutely essential. </p>\n<p>If you look at the <a href=\"https://birdnet.cornell.edu\" target=\"_blank\">BirdNET</a> app as an example, it makes predictions for species that occur in North America and Europe, but it then uses geographic and date filters to figure out which ones are reasonable possibilities, so that the end user of the app only sees these sensible possibilities. This is explained in <a href=\"https://youtu.be/MQHunTLt1TI\" target=\"_blank\">an excellent recorded webinar of an interview of Stefan Kahl </a>. </p>\n<p><a href=\"https://ebird.org/home\" target=\"_blank\">eBird</a> has a large collection of citizen-science observations that do a pretty good job of documenting the when and where of bird species occurrences. </p>\n<p>eBird has several data products:<br>\n1) Raw eBird data. To access the data, a written request is required, and data use comes with some restrictions. It’s not entirely clear whether the raw dataset would be permissible in this competition, but I don’t think it is necessary because of other options. </p>\n<p>2) “<a href=\"https://ebird.org/science/status-and-trends/\" target=\"_blank\">Status and Trends</a>” maps and data products use species distribution modeling to predict species occurrence at a fine spatial and temporal resolution. These are amazing, but they aren’t available for all species. The visualizations are available online, but I’m not sure about download options for the data products.</p>\n<p>3) Finally, a very practical option: eBird bar charts. These bar charts summarize bird occurrence frequency by time of year for locations that include either hotspots (places that have many reports from birders) or regions (countries, provinces, states, counties, etc.). </p>\n<p>Here’s a quick guide on how to access the eBird Bar Chart data:</p>\n<ul>\n<li>To complete all of the steps, a Cornell Lab account is required, but this is free and easy from the Get Started link on the main eBird page.</li>\n<li>Go to<a href=\"https://ebird.org/explore\" target=\"_blank\"> ebird.org/explore</a> and click on Bar Charts. </li>\n<li>Select a location; this might be a political region, or it might be a hotspot. Selecting multiple hotspots is a possibility (and the approach I lean towards; see <a href=\"https://ebird.org/hotspots\" target=\"_blank\">https://ebird.org/hotspots</a> to figure out which ones might be appropriate). </li>\n<li>This will bring up a page with graphics for occurrence frequency by time for the full year for all relevant species. At the top of the page, you can “Change Date Range” to narrow down the time window. The output from this is a little confusing, because it will still show the bird occurrence frequency data for the full year, but it will indeed filter to only the species with records in the requested date range.</li>\n<li>At the bottom of the page, there is a link to “Download Histogram Data”. The output has a slightly awkward format, but it is just basic data wrangling to turn it into a usable data product. </li>\n</ul>",
      "rawMarkdown": "From a birder’s perspective, taking location and date into account is absolutely essential. \n\nIf you look at the [BirdNET](https://birdnet.cornell.edu) app as an example, it makes predictions for species that occur in North America and Europe, but it then uses geographic and date filters to figure out which ones are reasonable possibilities, so that the end user of the app only sees these sensible possibilities. This is explained in [an excellent recorded webinar of an interview of Stefan Kahl ](https://youtu.be/MQHunTLt1TI). \n\n[eBird](https://ebird.org/home) has a large collection of citizen-science observations that do a pretty good job of documenting the when and where of bird species occurrences. \n\neBird has several data products:\n1) Raw eBird data. To access the data, a written request is required, and data use comes with some restrictions. It’s not entirely clear whether the raw dataset would be permissible in this competition, but I don’t think it is necessary because of other options. \n\n2) “[Status and Trends](https://ebird.org/science/status-and-trends/)” maps and data products use species distribution modeling to predict species occurrence at a fine spatial and temporal resolution. These are amazing, but they aren’t available for all species. The visualizations are available online, but I’m not sure about download options for the data products.\n\n3) Finally, a very practical option: eBird bar charts. These bar charts summarize bird occurrence frequency by time of year for locations that include either hotspots (places that have many reports from birders) or regions (countries, provinces, states, counties, etc.). \n\nHere’s a quick guide on how to access the eBird Bar Chart data:\n* To complete all of the steps, a Cornell Lab account is required, but this is free and easy from the Get Started link on the main eBird page.\n* Go to[ ebird.org/explore](https://ebird.org/explore) and click on Bar Charts. \n* Select a location; this might be a political region, or it might be a hotspot. Selecting multiple hotspots is a possibility (and the approach I lean towards; see https://ebird.org/hotspots to figure out which ones might be appropriate). \n* This will bring up a page with graphics for occurrence frequency by time for the full year for all relevant species. At the top of the page, you can “Change Date Range” to narrow down the time window. The output from this is a little confusing, because it will still show the bird occurrence frequency data for the full year, but it will indeed filter to only the species with records in the requested date range.\n* At the bottom of the page, there is a link to “Download Histogram Data”. The output has a slightly awkward format, but it is just basic data wrangling to turn it into a usable data product.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1309008,
      "author_name": "jmreuter",
      "author_url": "",
      "post_date": "05/15/2021 15:33:58",
      "content": "<p>From a birder’s perspective, taking location and date into account is absolutely essential. </p>\n<p>If you look at the <a href=\"https://birdnet.cornell.edu\" target=\"_blank\">BirdNET</a> app as an example, it makes predictions for species that occur in North America and Europe, but it then uses geographic and date filters to figure out which ones are reasonable possibilities, so that the end user of the app only sees these sensible possibilities. This is explained in <a href=\"https://youtu.be/MQHunTLt1TI\" target=\"_blank\">an excellent recorded webinar of an interview of Stefan Kahl </a>. </p>\n<p><a href=\"https://ebird.org/home\" target=\"_blank\">eBird</a> has a large collection of citizen-science observations that do a pretty good job of documenting the when and where of bird species occurrences. </p>\n<p>eBird has several data products:<br>\n1) Raw eBird data. To access the data, a written request is required, and data use comes with some restrictions. It’s not entirely clear whether the raw dataset would be permissible in this competition, but I don’t think it is necessary because of other options. </p>\n<p>2) “<a href=\"https://ebird.org/science/status-and-trends/\" target=\"_blank\">Status and Trends</a>” maps and data products use species distribution modeling to predict species occurrence at a fine spatial and temporal resolution. These are amazing, but they aren’t available for all species. The visualizations are available online, but I’m not sure about download options for the data products.</p>\n<p>3) Finally, a very practical option: eBird bar charts. These bar charts summarize bird occurrence frequency by time of year for locations that include either hotspots (places that have many reports from birders) or regions (countries, provinces, states, counties, etc.). </p>\n<p>Here’s a quick guide on how to access the eBird Bar Chart data:</p>\n<ul>\n<li>To complete all of the steps, a Cornell Lab account is required, but this is free and easy from the Get Started link on the main eBird page.</li>\n<li>Go to<a href=\"https://ebird.org/explore\" target=\"_blank\"> ebird.org/explore</a> and click on Bar Charts. </li>\n<li>Select a location; this might be a political region, or it might be a hotspot. Selecting multiple hotspots is a possibility (and the approach I lean towards; see <a href=\"https://ebird.org/hotspots\" target=\"_blank\">https://ebird.org/hotspots</a> to figure out which ones might be appropriate). </li>\n<li>This will bring up a page with graphics for occurrence frequency by time for the full year for all relevant species. At the top of the page, you can “Change Date Range” to narrow down the time window. The output from this is a little confusing, because it will still show the bird occurrence frequency data for the full year, but it will indeed filter to only the species with records in the requested date range.</li>\n<li>At the bottom of the page, there is a link to “Download Histogram Data”. The output has a slightly awkward format, but it is just basic data wrangling to turn it into a usable data product. </li>\n</ul>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1308093": "I would like to know ... if someone wishes to comment on how are you addressing the following points ... and if in any way anyone believes that it is worth taking them into account....\n\n- test_soundscapes **approximately** 80 recordings  **roughly** 10 minutes long \n- file names include the **date** the recording was taken\n- text files with the **name** and **approximate coordinates** of the recording location\n- **date**s the test set soundscapes were recorded",
    "1309008": "From a birder’s perspective, taking location and date into account is absolutely essential. \n\nIf you look at the [BirdNET](https://birdnet.cornell.edu) app as an example, it makes predictions for species that occur in North America and Europe, but it then uses geographic and date filters to figure out which ones are reasonable possibilities, so that the end user of the app only sees these sensible possibilities. This is explained in [an excellent recorded webinar of an interview of Stefan Kahl ](https://youtu.be/MQHunTLt1TI). \n\n[eBird](https://ebird.org/home) has a large collection of citizen-science observations that do a pretty good job of documenting the when and where of bird species occurrences. \n\neBird has several data products:\n1) Raw eBird data. To access the data, a written request is required, and data use comes with some restrictions. It’s not entirely clear whether the raw dataset would be permissible in this competition, but I don’t think it is necessary because of other options. \n\n2) “[Status and Trends](https://ebird.org/science/status-and-trends/)” maps and data products use species distribution modeling to predict species occurrence at a fine spatial and temporal resolution. These are amazing, but they aren’t available for all species. The visualizations are available online, but I’m not sure about download options for the data products.\n\n3) Finally, a very practical option: eBird bar charts. These bar charts summarize bird occurrence frequency by time of year for locations that include either hotspots (places that have many reports from birders) or regions (countries, provinces, states, counties, etc.). \n\nHere’s a quick guide on how to access the eBird Bar Chart data:\n* To complete all of the steps, a Cornell Lab account is required, but this is free and easy from the Get Started link on the main eBird page.\n* Go to[ ebird.org/explore](https://ebird.org/explore) and click on Bar Charts. \n* Select a location; this might be a political region, or it might be a hotspot. Selecting multiple hotspots is a possibility (and the approach I lean towards; see https://ebird.org/hotspots to figure out which ones might be appropriate). \n* This will bring up a page with graphics for occurrence frequency by time for the full year for all relevant species. At the top of the page, you can “Change Date Range” to narrow down the time window. The output from this is a little confusing, because it will still show the bird occurrence frequency data for the full year, but it will indeed filter to only the species with records in the requested date range.\n* At the bottom of the page, there is a link to “Download Histogram Data”. The output has a slightly awkward format, but it is just basic data wrangling to turn it into a usable data product."
  },
  "source": "meta"
}