{
  "id": 256530,
  "title": "I gave up this contest due to \"randomness\" of timestamps",
  "url": "/competitions/google-smartphone-decimeter-challenge/discussion/256530",
  "author_name": "Hirokazu Tamura",
  "post_date": "2021-08-02T05:58:41.830000",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>[I apologize first that my English is not accurate enough, because I am not a native speaker/writer of English…]</p>\n<p>I had wanted to submit my estimation, but I gave up now, due to \"randomness\" of timestamp which <br>\nI could not solve this problem at interpolation and extrapolation process.</p>\n<p>e.g.</p>\n<ol>\n<li>I could not find any data at GPS time \"1281390074444\" in Pixel4_derived.csv file which is in \"2020-08-13-US-MTV-1\" folder, even considering that the timestamp in derived.csv is delayed 1 second. This timestamp is contained both in baseline and sample of submission.</li>\n<li>At the same file, GPS time is totally uneven, like \"1281392025444, 1281392053444, 1281392073444, 1281392074444\" even though the derived data only has data at \"1281392025444\" and \"1281392080444\". Nearby data at that data seems to have same problem. Therefore, I could not reproduce timestamp from the interpolation.</li>\n<li>Also I saw some uneven timestamp in other data, which might make interpolation failed and barrier of reproduction of timestamps.</li>\n</ol>\n<p>I considered \"timestamp should be even\", but fact is not.<br>\nI totally misunderstood this data set, and spoiled a couple of weeks of work.<br>\nI think I am totally not worthy to enter all these Kaggle competition.<br>\nWhat I all can do is only to disappear (or train myself again, but training myself will not happen, I think).</p>",
  "messages": [
    {
      "id": 1408203,
      "postDate": "2021-08-02T11:23:01.977Z",
      "content": "<p>That's tough.  </p>",
      "rawMarkdown": "That's tough.  ",
      "votes": 1
    },
    {
      "id": 1408189,
      "postDate": "2021-08-02T11:10:18.090Z",
      "content": "<p>See you space cowboy</p>",
      "rawMarkdown": "See you space cowboy",
      "votes": 1
    },
    {
      "id": 1407754,
      "postDate": "2021-08-02T05:58:41.830Z",
      "content": "<p>[I apologize first that my English is not accurate enough, because I am not a native speaker/writer of English…]</p>\n<p>I had wanted to submit my estimation, but I gave up now, due to \"randomness\" of timestamp which <br>\nI could not solve this problem at interpolation and extrapolation process.</p>\n<p>e.g.</p>\n<ol>\n<li>I could not find any data at GPS time \"1281390074444\" in Pixel4_derived.csv file which is in \"2020-08-13-US-MTV-1\" folder, even considering that the timestamp in derived.csv is delayed 1 second. This timestamp is contained both in baseline and sample of submission.</li>\n<li>At the same file, GPS time is totally uneven, like \"1281392025444, 1281392053444, 1281392073444, 1281392074444\" even though the derived data only has data at \"1281392025444\" and \"1281392080444\". Nearby data at that data seems to have same problem. Therefore, I could not reproduce timestamp from the interpolation.</li>\n<li>Also I saw some uneven timestamp in other data, which might make interpolation failed and barrier of reproduction of timestamps.</li>\n</ol>\n<p>I considered \"timestamp should be even\", but fact is not.<br>\nI totally misunderstood this data set, and spoiled a couple of weeks of work.<br>\nI think I am totally not worthy to enter all these Kaggle competition.<br>\nWhat I all can do is only to disappear (or train myself again, but training myself will not happen, I think).</p>",
      "rawMarkdown": "[I apologize first that my English is not accurate enough, because I am not a native speaker/writer of English...]\n\nI had wanted to submit my estimation, but I gave up now, due to \"randomness\" of timestamp which \nI could not solve this problem at interpolation and extrapolation process.\n\ne.g.\n1. I could not find any data at GPS time \"1281390074444\" in Pixel4_derived.csv file which is in \"2020-08-13-US-MTV-1\" folder, even considering that the timestamp in derived.csv is delayed 1 second. This timestamp is contained both in baseline and sample of submission.\n2. At the same file, GPS time is totally uneven, like \"1281392025444, 1281392053444, 1281392073444, 1281392074444\" even though the derived data only has data at \"1281392025444\" and \"1281392080444\". Nearby data at that data seems to have same problem. Therefore, I could not reproduce timestamp from the interpolation.\n3. Also I saw some uneven timestamp in other data, which might make interpolation failed and barrier of reproduction of timestamps.\n\nI considered \"timestamp should be even\", but fact is not.\nI totally misunderstood this data set, and spoiled a couple of weeks of work.\nI think I am totally not worthy to enter all these Kaggle competition.\nWhat I all can do is only to disappear (or train myself again, but training myself will not happen, I think)."
    },
    {
      "id": 1419087,
      "postDate": "2021-08-03T02:03:56.477Z",
      "content": "<p>I have a little timestamp snippet </p>\n<pre><code>def timeBlocks_ind(v1, v2, xoffset=0):        \n\n    stamp0 = v2[0];    \n    v1x = np.round((v1 - stamp0)/1000.0) + xoffset;\n    v2x = np.round((v2 - stamp0)/1000.0);    \n    r1 = np.zeros(v1x.shape[0], int);\n    r2 = np.zeros(v2x.shape[0], int);    \n\n    for i in range(v1x.shape[0]):\n        if v1x[i] in v2x:\n            r1[i] = 1;\n    for i in range(v2x.shape[0]):\n        if v2x[i] in v1x:\n            r2[i] = 1;\n    return [r1, r2];\n</code></pre>\n<p>iterable <br>\n<code>for i,j in zip(np.arange(r[0].shape)[r[0]==1], np.arange(r[1].shape)[r[1]==1]):</code></p>",
      "rawMarkdown": "I have a little timestamp snippet \n\n```\ndef timeBlocks_ind(v1, v2, xoffset=0):        \n\n    stamp0 = v2[0];    \n    v1x = np.round((v1 - stamp0)/1000.0) + xoffset;\n    v2x = np.round((v2 - stamp0)/1000.0);    \n    r1 = np.zeros(v1x.shape[0], int);\n    r2 = np.zeros(v2x.shape[0], int);    \n    \n    for i in range(v1x.shape[0]):\n        if v1x[i] in v2x:\n            r1[i] = 1;\n    for i in range(v2x.shape[0]):\n        if v2x[i] in v1x:\n            r2[i] = 1;\n    return [r1, r2];\n```\n\niterable \n`for i,j in zip(np.arange(r[0].shape)[r[0]==1], np.arange(r[1].shape)[r[1]==1]):`"
    },
    {
      "id": 1451567,
      "postDate": "2021-08-05T11:32:54.483Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 1408203,
      "author_name": "T88",
      "author_url": "",
      "post_date": "2021-08-02T11:23:01.977000",
      "content": "<p>That's tough.  </p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 1408189,
      "author_name": "majoraregalia",
      "author_url": "",
      "post_date": "2021-08-02T11:10:18.090000",
      "content": "<p>See you space cowboy</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 1419087,
      "author_name": "n3n7i",
      "author_url": "",
      "post_date": "2021-08-03T02:03:56.477000",
      "content": "<p>I have a little timestamp snippet </p>\n<pre><code>def timeBlocks_ind(v1, v2, xoffset=0):        \n\n    stamp0 = v2[0];    \n    v1x = np.round((v1 - stamp0)/1000.0) + xoffset;\n    v2x = np.round((v2 - stamp0)/1000.0);    \n    r1 = np.zeros(v1x.shape[0], int);\n    r2 = np.zeros(v2x.shape[0], int);    \n\n    for i in range(v1x.shape[0]):\n        if v1x[i] in v2x:\n            r1[i] = 1;\n    for i in range(v2x.shape[0]):\n        if v2x[i] in v1x:\n            r2[i] = 1;\n    return [r1, r2];\n</code></pre>\n<p>iterable <br>\n<code>for i,j in zip(np.arange(r[0].shape)[r[0]==1], np.arange(r[1].shape)[r[1]==1]):</code></p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 1451567,
      "author_name": "",
      "author_url": "",
      "post_date": "2021-08-05T11:32:54.483000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1408203": "That's tough.  ",
    "1408189": "See you space cowboy",
    "1407754": "[I apologize first that my English is not accurate enough, because I am not a native speaker/writer of English...]\n\nI had wanted to submit my estimation, but I gave up now, due to \"randomness\" of timestamp which \nI could not solve this problem at interpolation and extrapolation process.\n\ne.g.\n1. I could not find any data at GPS time \"1281390074444\" in Pixel4_derived.csv file which is in \"2020-08-13-US-MTV-1\" folder, even considering that the timestamp in derived.csv is delayed 1 second. This timestamp is contained both in baseline and sample of submission.\n2. At the same file, GPS time is totally uneven, like \"1281392025444, 1281392053444, 1281392073444, 1281392074444\" even though the derived data only has data at \"1281392025444\" and \"1281392080444\". Nearby data at that data seems to have same problem. Therefore, I could not reproduce timestamp from the interpolation.\n3. Also I saw some uneven timestamp in other data, which might make interpolation failed and barrier of reproduction of timestamps.\n\nI considered \"timestamp should be even\", but fact is not.\nI totally misunderstood this data set, and spoiled a couple of weeks of work.\nI think I am totally not worthy to enter all these Kaggle competition.\nWhat I all can do is only to disappear (or train myself again, but training myself will not happen, I think).",
    "1419087": "I have a little timestamp snippet \n\n```\ndef timeBlocks_ind(v1, v2, xoffset=0):        \n\n    stamp0 = v2[0];    \n    v1x = np.round((v1 - stamp0)/1000.0) + xoffset;\n    v2x = np.round((v2 - stamp0)/1000.0);    \n    r1 = np.zeros(v1x.shape[0], int);\n    r2 = np.zeros(v2x.shape[0], int);    \n    \n    for i in range(v1x.shape[0]):\n        if v1x[i] in v2x:\n            r1[i] = 1;\n    for i in range(v2x.shape[0]):\n        if v2x[i] in v1x:\n            r2[i] = 1;\n    return [r1, r2];\n```\n\niterable \n`for i,j in zip(np.arange(r[0].shape)[r[0]==1], np.arange(r[1].shape)[r[1]==1]):`",
    "1451567": ""
  }
}