{
  "id": 482600,
  "title": "Features i.e. peaks",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/482600",
  "author_name": "Iulian Cozma",
  "post_date": "2024-03-08T12:20:54.329000",
  "votes": 2,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I am looking for some meaningful features for my <a href=\"https://www.kaggle.com/code/icozma/hms-pyspark-exercise\" target=\"_blank\">notebook</a>. Now the features are based on something like bellow, which I guess can look better:&nbsp; </p>\n<pre><code>np.mean(x)  \nnp.std(x)  \nnp.var(x)    \nnp.median(x)   \nnp.(x))     \nnp.(x))   \nnp.(x) - np.(x)  \n</code></pre>\n<p>Curious&nbsp;if you explored others </p>\n<p>i.e. peaks related features <br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F362466%2F55e955d479d8773ef2096ba50365a85b%2Foutput.png?generation=1709899937422812&amp;alt=media\"></p>\n<p>or maybe some domain specific metrics. Some hint will probably keep my mind busy on how to implement them :-)</p>\n<p>Many thanks</p>",
  "messages": [
    {
      "id": 2687275,
      "postDate": "2024-03-08T12:20:54.330Z",
      "content": "<p>I am looking for some meaningful features for my <a href=\"https://www.kaggle.com/code/icozma/hms-pyspark-exercise\" target=\"_blank\">notebook</a>. Now the features are based on something like bellow, which I guess can look better:&nbsp; </p>\n<pre><code>np.mean(x)  \nnp.std(x)  \nnp.var(x)    \nnp.median(x)   \nnp.(x))     \nnp.(x))   \nnp.(x) - np.(x)  \n</code></pre>\n<p>Curious&nbsp;if you explored others </p>\n<p>i.e. peaks related features <br>\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F362466%2F55e955d479d8773ef2096ba50365a85b%2Foutput.png?generation=1709899937422812&amp;alt=media\"></p>\n<p>or maybe some domain specific metrics. Some hint will probably keep my mind busy on how to implement them :-)</p>\n<p>Many thanks</p>",
      "rawMarkdown": "I am looking for some meaningful features for my [notebook](https://www.kaggle.com/code/icozma/hms-pyspark-exercise). Now the features are based on something like bellow, which I guess can look better:  \n```python\nnp.mean(x)  \nnp.std(x)  \nnp.var(x)    \nnp.median(x)   \nnp.max(x))     \nnp.min(x))   \nnp.max(x) - np.min(x)  \n```\nCurious if you explored others \n\ni.e. peaks related features \n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F362466%2F55e955d479d8773ef2096ba50365a85b%2Foutput.png?generation=1709899937422812&alt=media)\n\nor maybe some domain specific metrics. Some hint will probably keep my mind busy on how to implement them :-)\n\nMany thanks\n",
      "votes": 2
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "2687275": "I am looking for some meaningful features for my [notebook](https://www.kaggle.com/code/icozma/hms-pyspark-exercise). Now the features are based on something like bellow, which I guess can look better:  \n```python\nnp.mean(x)  \nnp.std(x)  \nnp.var(x)    \nnp.median(x)   \nnp.max(x))     \nnp.min(x))   \nnp.max(x) - np.min(x)  \n```\nCurious if you explored others \n\ni.e. peaks related features \n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F362466%2F55e955d479d8773ef2096ba50365a85b%2Foutput.png?generation=1709899937422812&alt=media)\n\nor maybe some domain specific metrics. Some hint will probably keep my mind busy on how to implement them :-)\n\nMany thanks\n"
  }
}