{
  "id": 77655,
  "title": "Are you using subsampling or derivative functions ?",
  "url": "/competitions/LANL-Earthquake-Prediction/discussion/77655",
  "author_name": "Gabriel Preda",
  "post_date": "2019-01-15T09:23:37.314000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Is anyone using sub-sampling for the segments? I know (from the public Kernels) that there are several solutions calculating features by aggregating over the segment (calculating various aggregation functions). \nBut does anyone use instead random sampling of the segment  (or a combination of aggregated features and sampled values) ?\nIs anyone using derivatives of the signal instead of averaging ?</p>",
  "messages": [
    {
      "id": 456172,
      "postDate": "2019-01-15T09:23:37.313Z",
      "content": "<p>Is anyone using sub-sampling for the segments? I know (from the public Kernels) that there are several solutions calculating features by aggregating over the segment (calculating various aggregation functions). \nBut does anyone use instead random sampling of the segment  (or a combination of aggregated features and sampled values) ?\nIs anyone using derivatives of the signal instead of averaging ?</p>",
      "rawMarkdown": "Is anyone using sub-sampling for the segments? I know (from the public Kernels) that there are several solutions calculating features by aggregating over the segment (calculating various aggregation functions). \nBut does anyone use instead random sampling of the segment  (or a combination of aggregated features and sampled values) ?\nIs anyone using derivatives of the signal instead of averaging ?",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "456172": "Is anyone using sub-sampling for the segments? I know (from the public Kernels) that there are several solutions calculating features by aggregating over the segment (calculating various aggregation functions). \nBut does anyone use instead random sampling of the segment  (or a combination of aggregated features and sampled values) ?\nIs anyone using derivatives of the signal instead of averaging ?"
  }
}