{
  "id": 5784,
  "title": "Log base for srch_query_affinity_score",
  "url": "/competitions/expedia-personalized-sort/discussion/5784",
  "author_name": "",
  "post_date": "2013-09-17T09:01:49.807Z",
  "votes": null,
  "comment_count": 9,
  "views": 2464,
  "content": "<p>Found on site: &quot;The log of the probability a hotel will be clicked on in Internet searches (hence the values are negative) A null signifies there are no data (i.e. hotel did not register in any searches)&quot;</p>\n<p>Which is the base for taking log value?</p>\n<p>striki</p>",
  "messages": [
    {
      "id": "31029",
      "postDate": "09/17/2013 09:01:49",
      "content": "<p>Found on site: &quot;The log of the probability a hotel will be clicked on in Internet searches (hence the values are negative) A null signifies there are no data (i.e. hotel did not register in any searches)&quot;</p>\n<p>Which is the base for taking log value?</p>\n<p>striki</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31039",
      "postDate": "09/17/2013 11:32:26",
      "content": "<p>Additionally, what is the base for &quot;prop_log_historical_price&quot;&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31042",
      "postDate": "09/17/2013 12:10:50",
      "content": "<p>prop_log_historical_price is obtained using natural logarithm. </p>\n<p>I'll check srch_query_affinity_score and I'll let you know.</p>\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31058",
      "postDate": "09/17/2013 18:43:59",
      "content": "<p>srch_query_affinity_score is obtained using logarithm of base 10.</p>\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31331",
      "postDate": "09/22/2013 07:21:07",
      "content": "<p>Does higher value of this parameter for some property&nbsp;means that there is bigger probability that thath property will be clicked when appears in exedia searches in test results?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31336",
      "postDate": "09/22/2013 14:52:31",
      "content": "<p>yes</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31859",
      "postDate": "10/01/2013 12:59:10",
      "content": "<p>Follow up question @Adam - I noticed that the&nbsp;srch_query_affinity_score is showing up greater than 1 (with negative signs of-course) .You mentioned that it is natural log of probabilities(in which case it should be log(a number between 0 to 1, both included), how are the values showing up as -21.. , -81.. ? (Or is it multiplied by 10 straightforward post this step ?)</p>\n<p>PS : It is quite possible that the format of this data field may have messed up post my my sampling, so just clarifying, if I misunderstood something.&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31861",
      "postDate": "10/01/2013 13:11:33",
      "content": "<p>This is how the (natural) logarithm function works, e.g.:</p>\n<p>log(e^-21) = -21</p>\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31862",
      "postDate": "10/01/2013 13:28:25",
      "content": "<p>@Adam - Not really. It was my mistake.&nbsp;<strong>srch_query_affinity_score is not the natural log, but the base 10</strong> (as you wrote above) - meaning -:</p>\n<p>math.log(x)/math.log(10)=-21 - where x is the&nbsp;probability a hotel will be clicked on in Internet searches</p>\n<p>==&gt;10 ** (-21*math.log(10)) &nbsp;{where ** &nbsp;= ^ in mathematical terms} &nbsp;- this gives me&nbsp;4.422960360259454e-49 , which is the correct prob. value. (Please) Nudge me if I am incorrect now.&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31863",
      "postDate": "10/01/2013 13:43:00",
      "content": "<p>yes, srch_query_affinity_score is obtained using logarithm of base 10; so log10(10^-21) = -21</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 31039,
      "author_name": "ektagrover1",
      "author_url": "",
      "post_date": "09/17/2013 11:32:26",
      "content": "<p>Additionally, what is the base for &quot;prop_log_historical_price&quot;&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31042,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "09/17/2013 12:10:50",
      "content": "<p>prop_log_historical_price is obtained using natural logarithm. </p>\n<p>I'll check srch_query_affinity_score and I'll let you know.</p>\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31058,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "09/17/2013 18:43:59",
      "content": "<p>srch_query_affinity_score is obtained using logarithm of base 10.</p>\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31331,
      "author_name": "striki",
      "author_url": "",
      "post_date": "09/22/2013 07:21:07",
      "content": "<p>Does higher value of this parameter for some property&nbsp;means that there is bigger probability that thath property will be clicked when appears in exedia searches in test results?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31336,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "09/22/2013 14:52:31",
      "content": "<p>yes</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31859,
      "author_name": "ektagrover1",
      "author_url": "",
      "post_date": "10/01/2013 12:59:10",
      "content": "<p>Follow up question @Adam - I noticed that the&nbsp;srch_query_affinity_score is showing up greater than 1 (with negative signs of-course) .You mentioned that it is natural log of probabilities(in which case it should be log(a number between 0 to 1, both included), how are the values showing up as -21.. , -81.. ? (Or is it multiplied by 10 straightforward post this step ?)</p>\n<p>PS : It is quite possible that the format of this data field may have messed up post my my sampling, so just clarifying, if I misunderstood something.&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31861,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "10/01/2013 13:11:33",
      "content": "<p>This is how the (natural) logarithm function works, e.g.:</p>\n<p>log(e^-21) = -21</p>\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31862,
      "author_name": "ektagrover1",
      "author_url": "",
      "post_date": "10/01/2013 13:28:25",
      "content": "<p>@Adam - Not really. It was my mistake.&nbsp;<strong>srch_query_affinity_score is not the natural log, but the base 10</strong> (as you wrote above) - meaning -:</p>\n<p>math.log(x)/math.log(10)=-21 - where x is the&nbsp;probability a hotel will be clicked on in Internet searches</p>\n<p>==&gt;10 ** (-21*math.log(10)) &nbsp;{where ** &nbsp;= ^ in mathematical terms} &nbsp;- this gives me&nbsp;4.422960360259454e-49 , which is the correct prob. value. (Please) Nudge me if I am incorrect now.&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31863,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "10/01/2013 13:43:00",
      "content": "<p>yes, srch_query_affinity_score is obtained using logarithm of base 10; so log10(10^-21) = -21</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "31029": "",
    "31039": "",
    "31042": "",
    "31058": "",
    "31331": "",
    "31336": "",
    "31859": "",
    "31861": "",
    "31862": "",
    "31863": ""
  },
  "source": "meta"
}