{
  "id": 294963,
  "title": "PCA optimal components may change based on what model your using",
  "url": "/competitions/cs5785-fall-2021-final/discussion/294963",
  "author_name": "",
  "post_date": "2021-12-13T18:03:23.671282600Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Just wanted to add on to the <a href=\"https://www.kaggle.com/c/cs5785-fall-2021-final/discussion/294677\" target=\"_blank\">PCA on ResNet features discussion</a>. When doing PCA I found that lower dimensionality worked better when working with the baseline Ridge Regression model given to us however, when I changed the ml model to a more complicated approach like Deep Learning I found that going higher in the dimensionality than the one I used for Ridge was actually better. Hope this helps!</p>",
  "messages": [
    {
      "id": "1616930",
      "postDate": "12/13/2021 18:03:23",
      "content": "<p>Just wanted to add on to the <a href=\"https://www.kaggle.com/c/cs5785-fall-2021-final/discussion/294677\" target=\"_blank\">PCA on ResNet features discussion</a>. When doing PCA I found that lower dimensionality worked better when working with the baseline Ridge Regression model given to us however, when I changed the ml model to a more complicated approach like Deep Learning I found that going higher in the dimensionality than the one I used for Ridge was actually better. Hope this helps!</p>",
      "rawMarkdown": "Just wanted to add on to the [PCA on ResNet features discussion](https://www.kaggle.com/c/cs5785-fall-2021-final/discussion/294677). When doing PCA I found that lower dimensionality worked better when working with the baseline Ridge Regression model given to us however, when I changed the ml model to a more complicated approach like Deep Learning I found that going higher in the dimensionality than the one I used for Ridge was actually better. Hope this helps!",
      "votes": null
    },
    {
      "id": "1617182",
      "postDate": "12/13/2021 23:13:20",
      "content": "<p>This is really helpful, thank you!</p>",
      "rawMarkdown": "This is really helpful, thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1617182,
      "author_name": "sofiathompson",
      "author_url": "",
      "post_date": "12/13/2021 23:13:20",
      "content": "<p>This is really helpful, thank you!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1616930": "Just wanted to add on to the [PCA on ResNet features discussion](https://www.kaggle.com/c/cs5785-fall-2021-final/discussion/294677). When doing PCA I found that lower dimensionality worked better when working with the baseline Ridge Regression model given to us however, when I changed the ml model to a more complicated approach like Deep Learning I found that going higher in the dimensionality than the one I used for Ridge was actually better. Hope this helps!",
    "1617182": "This is really helpful, thank you!"
  },
  "source": "meta"
}