{
  "id": 155011,
  "title": "A Look at the Data with t-SNE and UMAP dimensionality reduction algorithms",
  "url": "/competitions/siim-isic-melanoma-classification/discussion/155011",
  "author_name": "",
  "post_date": "2020-05-30T20:32:30.659866400Z",
  "votes": 4,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I have used the Rapids versions of t-SNE and UMAP to reduce the dimensionality of the dataset and see if we can get any insights form that way. The resulting 2D mappings are interesting, but due to the heavily imbalanced target, it may not be possible to get any deeper insights form this. You can take a look at it in the following kernel:</p>\n\n<p><a href=\"https://www.kaggle.com/tunguz/melanoma-tsne-and-umap-embeddings-with-rapids\">https://www.kaggle.com/tunguz/melanoma-tsne-and-umap-embeddings-with-rapids</a></p>\n\n<p><img src=\"https://pbs.twimg.com/media/EZStlRfXgAEhViQ?format=png&amp;name=small\" alt=\"mel-tsne\"></p>",
  "messages": [
    {
      "id": "868035",
      "postDate": "05/30/2020 20:32:30",
      "content": "<p>I have used the Rapids versions of t-SNE and UMAP to reduce the dimensionality of the dataset and see if we can get any insights form that way. The resulting 2D mappings are interesting, but due to the heavily imbalanced target, it may not be possible to get any deeper insights form this. You can take a look at it in the following kernel:</p>\n\n<p><a href=\"https://www.kaggle.com/tunguz/melanoma-tsne-and-umap-embeddings-with-rapids\">https://www.kaggle.com/tunguz/melanoma-tsne-and-umap-embeddings-with-rapids</a></p>\n\n<p><img src=\"https://pbs.twimg.com/media/EZStlRfXgAEhViQ?format=png&amp;name=small\" alt=\"mel-tsne\"></p>",
      "rawMarkdown": "I have used the Rapids versions of t-SNE and UMAP to reduce the dimensionality of the dataset and see if we can get any insights form that way. The resulting 2D mappings are interesting, but due to the heavily imbalanced target, it may not be possible to get any deeper insights form this. You can take a look at it in the following kernel:\n\nhttps://www.kaggle.com/tunguz/melanoma-tsne-and-umap-embeddings-with-rapids\n\n![mel-tsne](https://pbs.twimg.com/media/EZStlRfXgAEhViQ?format=png&amp;name=small)",
      "votes": null
    },
    {
      "id": "868862",
      "postDate": "05/31/2020 14:38:03",
      "content": "<p>I tried to install Rapids but i have in return the following error:</p>\n\n<p><code>cp: cannot stat '../input/rapids/rapids.0.13.0': No such file or directory\ntar (child): rapids.tar.gz: Cannot open: No such file or directory\ntar (child): Error is not recoverable: exiting now\ntar: Child returned status 2\ntar: Error is not recoverable: exiting now\ncp: cannot stat '/opt/conda/envs/rapids/lib/libxgboost.so': No such file or directory</code></p>\n\n<p>I used this to install Rapids:\n<code>import sys\n!cp ../input/rapids/rapids.0.13.0 /opt/conda/envs/rapids.tar.gz\n!cd /opt/conda/envs/ &amp;&amp; tar -xzvf rapids.tar.gz &gt; /dev/null\nsys.path = [\"/opt/conda/envs/rapids/lib/python3.6/site-packages\"] + sys.path\nsys.path = [\"/opt/conda/envs/rapids/lib/python3.6\"] + sys.path\nsys.path = [\"/opt/conda/envs/rapids/lib\"] + sys.path\n!cp /opt/conda/envs/rapids/lib/libxgboost.so /opt/conda/lib/</code></p>",
      "rawMarkdown": "I tried to install Rapids but i have in return the following error:\n\n`cp: cannot stat '../input/rapids/rapids.0.13.0': No such file or directory\ntar (child): rapids.tar.gz: Cannot open: No such file or directory\ntar (child): Error is not recoverable: exiting now\ntar: Child returned status 2\ntar: Error is not recoverable: exiting now\ncp: cannot stat '/opt/conda/envs/rapids/lib/libxgboost.so': No such file or directory`\n\nI used this to install Rapids:\n`import sys\n!cp ../input/rapids/rapids.0.13.0 /opt/conda/envs/rapids.tar.gz\n!cd /opt/conda/envs/ &amp;&amp; tar -xzvf rapids.tar.gz &gt; /dev/null\nsys.path = [\"/opt/conda/envs/rapids/lib/python3.6/site-packages\"] + sys.path\nsys.path = [\"/opt/conda/envs/rapids/lib/python3.6\"] + sys.path\nsys.path = [\"/opt/conda/envs/rapids/lib\"] + sys.path\n!cp /opt/conda/envs/rapids/lib/libxgboost.so /opt/conda/lib/`",
      "votes": null
    },
    {
      "id": "868957",
      "postDate": "05/31/2020 15:45:21",
      "content": "<p>Note that in Boyan's kernel the library is delivered as an input dataset (../input/rapids). So you'd have to add this dataset to your kernel for this method to work.</p>",
      "rawMarkdown": "Note that in Boyan's kernel the library is delivered as an input dataset (../input/rapids). So you'd have to add this dataset to your kernel for this method to work.",
      "votes": null
    },
    {
      "id": "869100",
      "postDate": "05/31/2020 18:30:38",
      "content": "<p>Thanks a lot, i do that and it worked.</p>",
      "rawMarkdown": "Thanks a lot, i do that and it worked.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 868862,
      "author_name": "hiramcho",
      "author_url": "",
      "post_date": "05/31/2020 14:38:03",
      "content": "<p>I tried to install Rapids but i have in return the following error:</p>\n\n<p><code>cp: cannot stat '../input/rapids/rapids.0.13.0': No such file or directory\ntar (child): rapids.tar.gz: Cannot open: No such file or directory\ntar (child): Error is not recoverable: exiting now\ntar: Child returned status 2\ntar: Error is not recoverable: exiting now\ncp: cannot stat '/opt/conda/envs/rapids/lib/libxgboost.so': No such file or directory</code></p>\n\n<p>I used this to install Rapids:\n<code>import sys\n!cp ../input/rapids/rapids.0.13.0 /opt/conda/envs/rapids.tar.gz\n!cd /opt/conda/envs/ &amp;&amp; tar -xzvf rapids.tar.gz &gt; /dev/null\nsys.path = [\"/opt/conda/envs/rapids/lib/python3.6/site-packages\"] + sys.path\nsys.path = [\"/opt/conda/envs/rapids/lib/python3.6\"] + sys.path\nsys.path = [\"/opt/conda/envs/rapids/lib\"] + sys.path\n!cp /opt/conda/envs/rapids/lib/libxgboost.so /opt/conda/lib/</code></p>",
      "votes": null,
      "replies": [
        {
          "id": 868957,
          "author_name": "hubwag",
          "author_url": "",
          "post_date": "05/31/2020 15:45:21",
          "content": "<p>Note that in Boyan's kernel the library is delivered as an input dataset (../input/rapids). So you'd have to add this dataset to your kernel for this method to work.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 869100,
          "author_name": "hiramcho",
          "author_url": "",
          "post_date": "05/31/2020 18:30:38",
          "content": "<p>Thanks a lot, i do that and it worked.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "868035": "I have used the Rapids versions of t-SNE and UMAP to reduce the dimensionality of the dataset and see if we can get any insights form that way. The resulting 2D mappings are interesting, but due to the heavily imbalanced target, it may not be possible to get any deeper insights form this. You can take a look at it in the following kernel:\n\nhttps://www.kaggle.com/tunguz/melanoma-tsne-and-umap-embeddings-with-rapids\n\n![mel-tsne](https://pbs.twimg.com/media/EZStlRfXgAEhViQ?format=png&amp;name=small)",
    "868862": "I tried to install Rapids but i have in return the following error:\n\n`cp: cannot stat '../input/rapids/rapids.0.13.0': No such file or directory\ntar (child): rapids.tar.gz: Cannot open: No such file or directory\ntar (child): Error is not recoverable: exiting now\ntar: Child returned status 2\ntar: Error is not recoverable: exiting now\ncp: cannot stat '/opt/conda/envs/rapids/lib/libxgboost.so': No such file or directory`\n\nI used this to install Rapids:\n`import sys\n!cp ../input/rapids/rapids.0.13.0 /opt/conda/envs/rapids.tar.gz\n!cd /opt/conda/envs/ &amp;&amp; tar -xzvf rapids.tar.gz &gt; /dev/null\nsys.path = [\"/opt/conda/envs/rapids/lib/python3.6/site-packages\"] + sys.path\nsys.path = [\"/opt/conda/envs/rapids/lib/python3.6\"] + sys.path\nsys.path = [\"/opt/conda/envs/rapids/lib\"] + sys.path\n!cp /opt/conda/envs/rapids/lib/libxgboost.so /opt/conda/lib/`",
    "868957": "Note that in Boyan's kernel the library is delivered as an input dataset (../input/rapids). So you'd have to add this dataset to your kernel for this method to work.",
    "869100": "Thanks a lot, i do that and it worked."
  },
  "source": "meta"
}