{
  "fetched_at": "2026-10-24T15:51:01Z",
  "source_url": "https://www.kaggle.com/competitions/dmse-2-1/data",
  "pages": [
    {
      "name": "data-description",
      "content": "<p>OPERA detector consists of several huge layers of special sensitive blocks *).</p>\r\n<p><img src=\"https://storage.googleapis.com/kaggle-media/competitions/inclass/6986/media/opera_2d.png\" alt=\"OPERA overall\" width=\"600\" height=\"346\" /></p>\r\n<p>Each layer is constructed as a wall of a special sandwich-like detection modules, called bricks *):</p>\r\n<p><img src=\"https://storage.googleapis.com/kaggle-media/competitions/inclass/6986/media/opera_ecc_scanning.png\" alt=\"\" width=\"400\" height=\"301\" /></p>\r\n<p>Such bricks interleave lead with photo emulsion. So as neutrino goes through the brick, it may interact with the lead and produce shower-like patterns, that are described by coordinates and angles of the initial particle and the collection of tracks that belong to the shower. The signal sample consists of simulated events. To obtain background sample the emulsion plates has been exposed to cosmic rays during creation, transportation, etc. Such traces which are mostly random is the main source of the background.</p>\r\n<h2>File descriptions</h2>\r\n<ul>\r\n<li><strong style=\"font-size: 1em;\">DS_2_train.csv</strong><span style=\"font-size: 1em;\"> - the training set</span></li>\r\n<li><strong>DS_2_test.csv</strong> - the test set</li>\r\n</ul>\r\n<h2>Data fields in training/test sets</h2>\r\n<p>training/test samples</p>\r\n<ul>\r\n<li><strong>Id</strong> - ID of a track</li>\r\n<li><strong>X</strong> - X coordinate of the track start</li>\r\n<li><strong>Y</strong> - Y coordinate of the track start</li>\r\n<li><strong>Z</strong> - Z coordinate of the track start</li>\r\n<li><strong>TX</strong> - angle of the track in XZ plane</li>\r\n<li><strong>TY</strong> - angle of the track in YZ plane</li>\r\n<li><strong>chi2</strong> - goodness of fit of the track line to the track hits</li>\r\n</ul>\r\n<p>there is <strong>signal</strong> field in the training set, which equals to <strong>1.0</strong> for signal event and to <strong>0.0</strong> otherwise.</p>\r\n<p>Training set also has field <strong>event_id</strong> either equals to -999, which means this track is a background or equals to shower ID (like in Episode 1) in case it is signal.</p>\r\n<p>Unlike Episode 1, there are no starting points of the showers. This makes the problem much more challenging, but much closer to the proposed SHIP detector setting.</p>\r\n<p>*) images are courtesy of OPERA collaboration.</p>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "<p>The data sample comes from OPERA (http://operaweb.lngs.infn.it) experiment that is well-known to find neutrinos using photo emulsion technique. Neutrinos themselves are invisible in the detector, but a very small fraction of them will interact with material in the detector and produce an electromagnetic shower in case of a scattered electron and also a hadronic shower in case of a scattered nucleus. The goal of the machine learning competition is to find traces of electromagnetic showers within a very dense background sample (see figure). Atomic electrons can also be scattered by light dark matter which is searched for by other experiments, including those at CERN, for example at SHiP (http://ship.web.cern.ch), which is going to use very similar neutrino detection techniques as the OPERA experiment.</p>\r\n<p><img src=\"https://storage.googleapis.com/kaggle-media/competitions/inclass/6986/media/hep_ann_signal.png\" alt=\"Background (blue) and electromagnetic shower (red) traces inside a detector module.\" width=\"600\" height=\"400\" /></p>\r\n<h2>Prize</h2>\r\n<p>We believe that the motivation is very important in such competitions, however being pure research-motivated project, we cannot offer much more than pure joy of giving a whirl to uncharted area of dark matter searches. Also we believe that sharing joy is really important, so winner of this competition will be invited to give a talk at the 12th SHiP collaboration meeting this year at CERN.</p>\r\n<h2>Acknowledgements</h2>\r\n<p>We would like to thank OPERA collaboration and Prof. Giovanni De Lellis, spokesperson of the collaboration, for sharing the data sample and supporting this challenge.</p>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "<p>Every submission is evaluated using standard ROC AUC formula that is both stable and well-known (and works well at kaggle inclass).</p>\r\n<h2>Submission Format</h2>\r\n<p><strong>For every track in the dataset</strong>, submission files should contain two columns: Id and Prediction separated by comma. Prediction is a positive value between 0 and 1.0 that indicates how likely the track belongs to a shower.</p>\r\n<p>The file should contain a header and have the following format:</p>\r\n<pre>Id,Prediction<br />1,0.1<br />2,0.4<br />3,0.1<br />4,0.9</pre>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "<!-- Begin Base Rules -->\r\n<h3>One account per participant</h3>\r\n<p>You cannot sign up to Kaggle from multiple accounts and therefore you cannot submit from multiple accounts.</p>\r\n<h3>No private sharing outside teams</h3>\r\n<p>Privately sharing code or data outside of teams is not permitted. It's okay to share code if made available to all participants on the forums.</p>\r\n<h3>Team Mergers</h3>\r\n<p>Team mergers are allowed in this competition.</p>\r\n<h3>External data</h3>\r\n<p>Use of external data is strictly prohibited.</p>\r\n<h3>Team Limits</h3>\r\n<p>There is no maximum team size.</p>\r\n<p>Team Limits</p>\r\n<p>There is no maximum team size.</p>\r\n<h3>Submission Limits</h3>\r\n<p>You may submit a maximum of 2 entries per day.</p>\r\n<p>You may select up to 2 final submissions for judging.</p>\r\n<h3>Competition Timeline</h3>\r\n<p>Start Date: <strong>30/08/2017</strong></p>\r\n<p>Merger Deadline: <strong>None</strong></p>\r\n<p>Entry Deadline: <strong>None</strong></p>\r\n<p>End Date: <strong>17/9/2017 12:00 AM UTC</strong></p>\r\n<!-- End Base Rules -->\r\n<p>The rules page is a bit special: it will auto-update the text based on the parameters of the competition. The rules you need to add here are any additional rules relating to dataset usage. You may also put grading information here.</p>",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}