{
  "id": 55858,
  "title": "Submission spends long time to upload a file",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/55858",
  "author_name": "",
  "post_date": "2018-05-02T13:22:34.093110600Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Can anyone tell why it happens? I experienced this since yesterday. </p>",
  "messages": [
    {
      "id": "322147",
      "postDate": "05/02/2018 13:22:34",
      "content": "<p>Can anyone tell why it happens? I experienced this since yesterday. </p>",
      "rawMarkdown": "Can anyone tell why it happens? I experienced this since yesterday.",
      "votes": null
    },
    {
      "id": "322208",
      "postDate": "05/02/2018 14:54:50",
      "content": "<p>You should compress it.  I recommend using 7z, but you can use gzip or zip too.</p>\n\n<p>Also saving ranked submission can save space, and accelerate upload, see this notebook by @Andy Harless: <a href=\"https://www.kaggle.com/aharless/smallification\">https://www.kaggle.com/aharless/smallification</a></p>\n\n<p>I don't get why you got downvoted.</p>",
      "rawMarkdown": "You should compress it.  I recommend using 7z, but you can use gzip or zip too.\n\nAlso saving ranked submission can save space, and accelerate upload, see this notebook by @Andy Harless: https://www.kaggle.com/aharless/smallification\n\nI don't get why you got downvoted.",
      "votes": null
    },
    {
      "id": "322524",
      "postDate": "05/03/2018 06:09:46",
      "content": "<p>if you are using linux, you can submit with the kaggle-cli via command line. </p>\n\n<p>All steps to install it and use it are here: <a href=\"https://github.com/floydwch/kaggle-cli\">https://github.com/floydwch/kaggle-cli</a></p>\n\n<p>before i used to run on GCP and then download to my machine and then upload in web-browser.</p>\n\n<p>Now i just do this: kg submit result_file.csv</p>",
      "rawMarkdown": "if you are using linux, you can submit with the kaggle-cli via command line. \n\nAll steps to install it and use it are here: https://github.com/floydwch/kaggle-cli\n\nbefore i used to run on GCP and then download to my machine and then upload in web-browser.\n\nNow i just do this: kg submit result_file.csv",
      "votes": null
    },
    {
      "id": "322564",
      "postDate": "05/03/2018 07:42:10",
      "content": "<p>I uploaded a file yesterday, using almost one night. Maybe my internet has something wrong.</p>",
      "rawMarkdown": "I uploaded a file yesterday, using almost one night. Maybe my internet has something wrong.",
      "votes": null
    },
    {
      "id": "322991",
      "postDate": "05/04/2018 05:02:26",
      "content": "<p>A simple way to compress the solution from pandas is:</p>\n\n<p>df.to_csv(\"solution.csv.gz\",index=False, compression='gzip')</p>\n\n<p>Another thing you should look at is the number of decimals your predictions have ... the more ... the bigger the file. I always round my predictions to 6th decimal. It has a minimum impact on the score.</p>",
      "rawMarkdown": "A simple way to compress the solution from pandas is:\n\ndf.to_csv(\"solution.csv.gz\",index=False, compression='gzip')\n\nAnother thing you should look at is the number of decimals your predictions have ... the more ... the bigger the file. I always round my predictions to 6th decimal. It has a minimum impact on the score.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 322208,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "05/02/2018 14:54:50",
      "content": "<p>You should compress it.  I recommend using 7z, but you can use gzip or zip too.</p>\n\n<p>Also saving ranked submission can save space, and accelerate upload, see this notebook by @Andy Harless: <a href=\"https://www.kaggle.com/aharless/smallification\">https://www.kaggle.com/aharless/smallification</a></p>\n\n<p>I don't get why you got downvoted.</p>",
      "votes": null,
      "replies": [
        {
          "id": 322564,
          "author_name": "jinmahkust",
          "author_url": "",
          "post_date": "05/03/2018 07:42:10",
          "content": "<p>I uploaded a file yesterday, using almost one night. Maybe my internet has something wrong.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 322991,
          "author_name": "profetul",
          "author_url": "",
          "post_date": "05/04/2018 05:02:26",
          "content": "<p>A simple way to compress the solution from pandas is:</p>\n\n<p>df.to_csv(\"solution.csv.gz\",index=False, compression='gzip')</p>\n\n<p>Another thing you should look at is the number of decimals your predictions have ... the more ... the bigger the file. I always round my predictions to 6th decimal. It has a minimum impact on the score.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 322524,
      "author_name": "mpsampat",
      "author_url": "",
      "post_date": "05/03/2018 06:09:46",
      "content": "<p>if you are using linux, you can submit with the kaggle-cli via command line. </p>\n\n<p>All steps to install it and use it are here: <a href=\"https://github.com/floydwch/kaggle-cli\">https://github.com/floydwch/kaggle-cli</a></p>\n\n<p>before i used to run on GCP and then download to my machine and then upload in web-browser.</p>\n\n<p>Now i just do this: kg submit result_file.csv</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "322147": "Can anyone tell why it happens? I experienced this since yesterday.",
    "322208": "You should compress it.  I recommend using 7z, but you can use gzip or zip too.\n\nAlso saving ranked submission can save space, and accelerate upload, see this notebook by @Andy Harless: https://www.kaggle.com/aharless/smallification\n\nI don't get why you got downvoted.",
    "322524": "if you are using linux, you can submit with the kaggle-cli via command line. \n\nAll steps to install it and use it are here: https://github.com/floydwch/kaggle-cli\n\nbefore i used to run on GCP and then download to my machine and then upload in web-browser.\n\nNow i just do this: kg submit result_file.csv",
    "322564": "I uploaded a file yesterday, using almost one night. Maybe my internet has something wrong.",
    "322991": "A simple way to compress the solution from pandas is:\n\ndf.to_csv(\"solution.csv.gz\",index=False, compression='gzip')\n\nAnother thing you should look at is the number of decimals your predictions have ... the more ... the bigger the file. I always round my predictions to 6th decimal. It has a minimum impact on the score."
  },
  "source": "meta"
}