{
  "fetched_at": "2026-10-19T19:06:12Z",
  "source_url": "https://www.kaggle.com/competitions/dsc6232-rwanda-summer2020-hw2/data",
  "pages": [
    {
      "name": "data-description",
      "content": "Since the satellite images are very high resolution and in color (i.e. each image is represented as three color matrices, red, green and blue), we have reduced the resolution and have converted the colors into gray-scale.\n\nYou may download the data and build your models on your computer. Or, if you choose to use `colab`, you do not need to download the data. Just open [this starter notebook](https://colab.research.google.com/drive/1KDcodDRvaWd6l1Q77pNJh29b-AY-QDjM?usp=sharing), the data will be loaded automatically in the starter notebook. ",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "<h2>Detecting Illegal Coffee Plantations and Deforestation</h2>\n\nFrom mountainous jungles to low-lying dry forests, people are utilizing and accessing the forests of the Greater Mekong like never before. As recently as the early 70s, forests blanketed three quarters of the Greater Mekong. The last 30 years have seen more than 30% of that forest cover disappear.\n\nFarmers are clearing forest cover for coffee plantations at an accelerating pace.\n\nForests landscapes and river systems in the Greater Mekong region are becoming increasingly fragmented. The remaining intact areas are increasingly smaller and isolated inside areas that have been heavily transformed.\n\nThis can make it difficult for some animals to find suitable habitat or to follow usual migration routes. It can also increase inbreeding, which in turn can increase the vulnerability of species to disease and other pressures.\n\nWhen habitats are isolated, species also become more vulnerable to climate change, as their ability to move to areas with more favourable conditions becomes increasingly limited. -- WWF\n\nSince illegal coffee plantations are often located deep in the forest, they are difficult to detect. Therefore, satellite images are now being used to detect these plantations. However, there are vast numbers of satellite images being collected and it is both expensive and inefficient to process these images using human experts.\n\nYour job today is to build a machine learning model that can automatically detect the presence of a coffee plantation in a satellite image. You are provided with a small training set of labeled images (the label is 1 if there is a coffee plantation in the image, and the label is 0 if there is not).",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "<p>Your submission will be evaluated based on the <b><a href=\"https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html\">macro-f1 score</a></b>.</p>\n\n<h2>Submission Format</h2>\n\nSubmit a file output.csv with the predictions , which has the index of each datapoint in the test set and the corresponding prediction, one per line.\n\nThe first few lines of output.csv should look like this:\n<pre>&lt;p&gt;\nid, label\n0, 0\n1, 0\n2, 1\n3, 0\n4, 1\n5, 0\n6, 0\n7, 0\n8, 0\n&lt;/p&gt;</pre>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "This homework is very free in format. It is meant for you to experience how it feels like to be a machine learning expert facing a real world dataset. Feel free to copy-paste and adapt any code you have been working on so far in class. Only using techniques seen in the lectures, you should be able to beat the baseline in the leaderboard. You are also welcome to try new approaches that we haven't seen in class. There is no limit to your imagination!\n\nYou are allowed to form teams of 1-2 persons. (2 persons is the maximum). While in the competition, respect your colleagues. You are welcome to help each other and work together, as long as you are actively contributing, and learning in the process.\n\nFinally, for those students that rank at the top of the Leaderboard, we will have special prices that will be announced afterwards. Apply yourself and have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}