{
  "id": 505987,
  "title": "why is the metic hacking so effective",
  "url": "/competitions/home-credit-credit-risk-model-stability/discussion/505987",
  "author_name": "",
  "post_date": "2024-05-20T04:03:22.047764300Z",
  "votes": 11,
  "comment_count": 2,
  "views": 0,
  "content": "<p>there is my vaild of train data, which caculate all gini score of weeks</p>\n<p>[ 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23<br>\n 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47<br>\n 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71<br>\n 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91] </p>\n<p>[0.7878092649407038, 0.7409836065573772, 0.7323351154478579, 0.7560198660916289, 0.7583125254382037, 0.7957219052662245, 0.7659847214489108, 0.7180548249459351, 0.7440476831848168, 0.6986866072284537, 0.7002742749565865, 0.7114947219501577, 0.7402645659928657, 0.7678253055355859, 0.691992296784707, 0.7274278570112835, 0.7491478635962583, 0.7399843837895677, 0.7221904687190634, 0.7158207394973188, 0.7451097482246611, 0.7727570270871791, 0.6002585649644472, 0.7075337597725657, 0.6600189933523266, 0.5755062547515377, 0.7023816674675638, 0.6739211618516385, 0.7290733568037431, 0.6831081904750131, 0.6700958611337839, 0.6681749032371211, 0.6927486647304477, 0.7026575613728268, 0.6917823726223067, 0.7024576970414262, 0.6923874609476559, 0.6760855946326505, 0.7090057369279963, 0.7147867591101675, 0.7179937736469415, 0.6918317410275467, 0.701393497320101, 0.6928641608303521, 0.745483632776861, 0.6659536060871822, 0.69543717572058, 0.7188257221704557, 0.6793031355437249, 0.6922307955696445, 0.692990986052765, 0.7236066308295008, 0.7052463331399226, 0.684660613905945, 0.7023675303723225, 0.6978351855608906, 0.713924326393585, 0.7003850450495706, 0.7026391856760212, 0.726351103114838, 0.7213735043001812, 0.6777640200294879, 0.7179007356945268, 0.7094818948087116, 0.6984401961212228, 0.7105279129070841, 0.6873034885714553, 0.6870252703027628, 0.7179629042195002, 0.719081436088236, 0.7198423826683298, 0.6868027575482505, 0.6721145082372129, 0.6808226907351496, 0.6735314446360063, 0.6882296983664207, 0.7115492477825609, 0.6694205028710292, 0.7049267309525056, 0.6641048417010158, 0.6869390954315775, 0.6512149310480717, 0.6477313974591652, 0.6438384713580965, 0.6821371284890596, 0.6354980051597987, 0.6713311528424069, 0.6810578145193162, 0.6570298177994678, 0.6667368299955188, 0.6583699416205084, 0.6417737985919807]</p>\n<p>the falling_rate  which is  min(0,a) of the mertric = -0.0007189610086093421<br>\nthe mertric score = <strong>0.6217188550575432</strong></p>\n<p>if i try this:<br>\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM&lt;1] - 0.01).clip(0)<br>\nthe mertric score = <strong>0.6250264840017028</strong></p>\n<p>if i try this:<br>\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM&lt;5] - 0.01).clip(0)<br>\nthe mertric score = <strong>0.64907607481513</strong></p>\n<p>if i try this:<br>\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM&lt;5] - 0.03).clip(0)<br>\nthe mertric score = <strong>0.6612691176139737</strong></p>\n<p>basically, u can easily boost your pb score , if u assume  u have a slightly decrease in your predictions by WEEK_NUM</p>",
  "messages": [
    {
      "id": "2824796",
      "postDate": "05/20/2024 04:03:22",
      "content": "<p>there is my vaild of train data, which caculate all gini score of weeks</p>\n<p>[ 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23<br>\n 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47<br>\n 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71<br>\n 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91] </p>\n<p>[0.7878092649407038, 0.7409836065573772, 0.7323351154478579, 0.7560198660916289, 0.7583125254382037, 0.7957219052662245, 0.7659847214489108, 0.7180548249459351, 0.7440476831848168, 0.6986866072284537, 0.7002742749565865, 0.7114947219501577, 0.7402645659928657, 0.7678253055355859, 0.691992296784707, 0.7274278570112835, 0.7491478635962583, 0.7399843837895677, 0.7221904687190634, 0.7158207394973188, 0.7451097482246611, 0.7727570270871791, 0.6002585649644472, 0.7075337597725657, 0.6600189933523266, 0.5755062547515377, 0.7023816674675638, 0.6739211618516385, 0.7290733568037431, 0.6831081904750131, 0.6700958611337839, 0.6681749032371211, 0.6927486647304477, 0.7026575613728268, 0.6917823726223067, 0.7024576970414262, 0.6923874609476559, 0.6760855946326505, 0.7090057369279963, 0.7147867591101675, 0.7179937736469415, 0.6918317410275467, 0.701393497320101, 0.6928641608303521, 0.745483632776861, 0.6659536060871822, 0.69543717572058, 0.7188257221704557, 0.6793031355437249, 0.6922307955696445, 0.692990986052765, 0.7236066308295008, 0.7052463331399226, 0.684660613905945, 0.7023675303723225, 0.6978351855608906, 0.713924326393585, 0.7003850450495706, 0.7026391856760212, 0.726351103114838, 0.7213735043001812, 0.6777640200294879, 0.7179007356945268, 0.7094818948087116, 0.6984401961212228, 0.7105279129070841, 0.6873034885714553, 0.6870252703027628, 0.7179629042195002, 0.719081436088236, 0.7198423826683298, 0.6868027575482505, 0.6721145082372129, 0.6808226907351496, 0.6735314446360063, 0.6882296983664207, 0.7115492477825609, 0.6694205028710292, 0.7049267309525056, 0.6641048417010158, 0.6869390954315775, 0.6512149310480717, 0.6477313974591652, 0.6438384713580965, 0.6821371284890596, 0.6354980051597987, 0.6713311528424069, 0.6810578145193162, 0.6570298177994678, 0.6667368299955188, 0.6583699416205084, 0.6417737985919807]</p>\n<p>the falling_rate  which is  min(0,a) of the mertric = -0.0007189610086093421<br>\nthe mertric score = <strong>0.6217188550575432</strong></p>\n<p>if i try this:<br>\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM&lt;1] - 0.01).clip(0)<br>\nthe mertric score = <strong>0.6250264840017028</strong></p>\n<p>if i try this:<br>\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM&lt;5] - 0.01).clip(0)<br>\nthe mertric score = <strong>0.64907607481513</strong></p>\n<p>if i try this:<br>\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM&lt;5] - 0.03).clip(0)<br>\nthe mertric score = <strong>0.6612691176139737</strong></p>\n<p>basically, u can easily boost your pb score , if u assume  u have a slightly decrease in your predictions by WEEK_NUM</p>",
      "rawMarkdown": "there is my vaild of train data, which caculate all gini score of weeks\n\n[ 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23\n 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47\n 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71\n 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91] \n\n[0.7878092649407038, 0.7409836065573772, 0.7323351154478579, 0.7560198660916289, 0.7583125254382037, 0.7957219052662245, 0.7659847214489108, 0.7180548249459351, 0.7440476831848168, 0.6986866072284537, 0.7002742749565865, 0.7114947219501577, 0.7402645659928657, 0.7678253055355859, 0.691992296784707, 0.7274278570112835, 0.7491478635962583, 0.7399843837895677, 0.7221904687190634, 0.7158207394973188, 0.7451097482246611, 0.7727570270871791, 0.6002585649644472, 0.7075337597725657, 0.6600189933523266, 0.5755062547515377, 0.7023816674675638, 0.6739211618516385, 0.7290733568037431, 0.6831081904750131, 0.6700958611337839, 0.6681749032371211, 0.6927486647304477, 0.7026575613728268, 0.6917823726223067, 0.7024576970414262, 0.6923874609476559, 0.6760855946326505, 0.7090057369279963, 0.7147867591101675, 0.7179937736469415, 0.6918317410275467, 0.701393497320101, 0.6928641608303521, 0.745483632776861, 0.6659536060871822, 0.69543717572058, 0.7188257221704557, 0.6793031355437249, 0.6922307955696445, 0.692990986052765, 0.7236066308295008, 0.7052463331399226, 0.684660613905945, 0.7023675303723225, 0.6978351855608906, 0.713924326393585, 0.7003850450495706, 0.7026391856760212, 0.726351103114838, 0.7213735043001812, 0.6777640200294879, 0.7179007356945268, 0.7094818948087116, 0.6984401961212228, 0.7105279129070841, 0.6873034885714553, 0.6870252703027628, 0.7179629042195002, 0.719081436088236, 0.7198423826683298, 0.6868027575482505, 0.6721145082372129, 0.6808226907351496, 0.6735314446360063, 0.6882296983664207, 0.7115492477825609, 0.6694205028710292, 0.7049267309525056, 0.6641048417010158, 0.6869390954315775, 0.6512149310480717, 0.6477313974591652, 0.6438384713580965, 0.6821371284890596, 0.6354980051597987, 0.6713311528424069, 0.6810578145193162, 0.6570298177994678, 0.6667368299955188, 0.6583699416205084, 0.6417737985919807]\n\nthe falling_rate  which is  min(0,a) of the mertric = -0.0007189610086093421\nthe mertric score = **0.6217188550575432**\n\nif i try this:\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM<1] - 0.01).clip(0)\nthe mertric score = **0.6250264840017028**\n\nif i try this:\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM<5] - 0.01).clip(0)\nthe mertric score = **0.64907607481513**\n\n\nif i try this:\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM<5] - 0.03).clip(0)\nthe mertric score = **0.6612691176139737**\n\n\nbasically, u can easily boost your pb score , if u assume  u have a slightly decrease in your predictions by WEEK_NUM",
      "votes": null
    },
    {
      "id": "2833705",
      "postDate": "05/24/2024 10:59:24",
      "content": "<p>May I ask how do you set the validation set? Thanks! </p>",
      "rawMarkdown": "May I ask how do you set the validation set? Thanks!",
      "votes": null
    },
    {
      "id": "2833796",
      "postDate": "05/24/2024 11:52:51",
      "content": "<p>Just simply k flod vaildation, by the way actually the WEEK_NUM is **reverse **for this test, so on my  validation set does not have neagtive falling_rate.</p>",
      "rawMarkdown": "Just simply k flod vaildation, by the way actually the WEEK_NUM is **reverse **for this test, so on my  validation set does not have neagtive falling_rate.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2833705,
      "author_name": "ray6767",
      "author_url": "",
      "post_date": "05/24/2024 10:59:24",
      "content": "<p>May I ask how do you set the validation set? Thanks! </p>",
      "votes": null,
      "replies": [
        {
          "id": 2833796,
          "author_name": "xianzwaikato",
          "author_url": "",
          "post_date": "05/24/2024 11:52:51",
          "content": "<p>Just simply k flod vaildation, by the way actually the WEEK_NUM is **reverse **for this test, so on my  validation set does not have neagtive falling_rate.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2824796": "there is my vaild of train data, which caculate all gini score of weeks\n\n[ 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23\n 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47\n 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71\n 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91] \n\n[0.7878092649407038, 0.7409836065573772, 0.7323351154478579, 0.7560198660916289, 0.7583125254382037, 0.7957219052662245, 0.7659847214489108, 0.7180548249459351, 0.7440476831848168, 0.6986866072284537, 0.7002742749565865, 0.7114947219501577, 0.7402645659928657, 0.7678253055355859, 0.691992296784707, 0.7274278570112835, 0.7491478635962583, 0.7399843837895677, 0.7221904687190634, 0.7158207394973188, 0.7451097482246611, 0.7727570270871791, 0.6002585649644472, 0.7075337597725657, 0.6600189933523266, 0.5755062547515377, 0.7023816674675638, 0.6739211618516385, 0.7290733568037431, 0.6831081904750131, 0.6700958611337839, 0.6681749032371211, 0.6927486647304477, 0.7026575613728268, 0.6917823726223067, 0.7024576970414262, 0.6923874609476559, 0.6760855946326505, 0.7090057369279963, 0.7147867591101675, 0.7179937736469415, 0.6918317410275467, 0.701393497320101, 0.6928641608303521, 0.745483632776861, 0.6659536060871822, 0.69543717572058, 0.7188257221704557, 0.6793031355437249, 0.6922307955696445, 0.692990986052765, 0.7236066308295008, 0.7052463331399226, 0.684660613905945, 0.7023675303723225, 0.6978351855608906, 0.713924326393585, 0.7003850450495706, 0.7026391856760212, 0.726351103114838, 0.7213735043001812, 0.6777640200294879, 0.7179007356945268, 0.7094818948087116, 0.6984401961212228, 0.7105279129070841, 0.6873034885714553, 0.6870252703027628, 0.7179629042195002, 0.719081436088236, 0.7198423826683298, 0.6868027575482505, 0.6721145082372129, 0.6808226907351496, 0.6735314446360063, 0.6882296983664207, 0.7115492477825609, 0.6694205028710292, 0.7049267309525056, 0.6641048417010158, 0.6869390954315775, 0.6512149310480717, 0.6477313974591652, 0.6438384713580965, 0.6821371284890596, 0.6354980051597987, 0.6713311528424069, 0.6810578145193162, 0.6570298177994678, 0.6667368299955188, 0.6583699416205084, 0.6417737985919807]\n\nthe falling_rate  which is  min(0,a) of the mertric = -0.0007189610086093421\nthe mertric score = **0.6217188550575432**\n\nif i try this:\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM<1] - 0.01).clip(0)\nthe mertric score = **0.6250264840017028**\n\nif i try this:\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM<5] - 0.01).clip(0)\nthe mertric score = **0.64907607481513**\n\n\nif i try this:\n (vaild[ \"vaild_pred\" ][ vaild.WEEK_NUM<5] - 0.03).clip(0)\nthe mertric score = **0.6612691176139737**\n\n\nbasically, u can easily boost your pb score , if u assume  u have a slightly decrease in your predictions by WEEK_NUM",
    "2833705": "May I ask how do you set the validation set? Thanks!",
    "2833796": "Just simply k flod vaildation, by the way actually the WEEK_NUM is **reverse **for this test, so on my  validation set does not have neagtive falling_rate."
  },
  "source": "meta"
}