{
  "id": 66749,
  "title": "Hyperparameter tuning",
  "url": "/competitions/ga-customer-revenue-prediction/discussion/66749",
  "author_name": "choco",
  "post_date": "2018-09-25T01:49:13.311000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hello kaggler. i tuned hyperparameter tuning using bayesian optimization like link(<a href=\"https://www.kaggle.com/chocozzz/bayesian-optimization-for-lightgbm/code?scriptVersionId=5962809\">https://www.kaggle.com/chocozzz/bayesian-optimization-for-lightgbm/code?scriptVersionId=5962809</a>)</p>\n\n<p>But the results is mystery. </p>\n\n<h2>Result</h2>\n\n<ul>\n<li><p><a href=\"https://www.kaggle.com/prashantkikani/rstudio-lgb-single-model-lb1-6607\">https://www.kaggle.com/prashantkikani/rstudio-lgb-single-model-lb1-6607</a> is \nLB: 15320 ==&gt; 15336\nCV: 1.60552376 ==&gt; 1.60596768</p></li>\n<li><p><a href=\"https://www.kaggle.com/youhanlee/stratified-sampling-for-regression-lb-1-6595\">https://www.kaggle.com/youhanlee/stratified-sampling-for-regression-lb-1-6595</a> is \nLB : 15232. ==&gt; 15324\nCV: 1.62005868 ==&gt; 1.60345295</p></li>\n</ul>\n\n<p>I understand that the first cv and lb are both off the wrong optimization than the existing tuning value. But the second time cv got better, lb dropped. Is this an overfitting? If I am overfitting, what should I do to solve this problem?</p>",
  "messages": [
    {
      "id": 393207,
      "postDate": "2018-09-25T01:49:13.310Z",
      "content": "<p>Hello kaggler. i tuned hyperparameter tuning using bayesian optimization like link(<a href=\"https://www.kaggle.com/chocozzz/bayesian-optimization-for-lightgbm/code?scriptVersionId=5962809\">https://www.kaggle.com/chocozzz/bayesian-optimization-for-lightgbm/code?scriptVersionId=5962809</a>)</p>\n\n<p>But the results is mystery. </p>\n\n<h2>Result</h2>\n\n<ul>\n<li><p><a href=\"https://www.kaggle.com/prashantkikani/rstudio-lgb-single-model-lb1-6607\">https://www.kaggle.com/prashantkikani/rstudio-lgb-single-model-lb1-6607</a> is \nLB: 15320 ==&gt; 15336\nCV: 1.60552376 ==&gt; 1.60596768</p></li>\n<li><p><a href=\"https://www.kaggle.com/youhanlee/stratified-sampling-for-regression-lb-1-6595\">https://www.kaggle.com/youhanlee/stratified-sampling-for-regression-lb-1-6595</a> is \nLB : 15232. ==&gt; 15324\nCV: 1.62005868 ==&gt; 1.60345295</p></li>\n</ul>\n\n<p>I understand that the first cv and lb are both off the wrong optimization than the existing tuning value. But the second time cv got better, lb dropped. Is this an overfitting? If I am overfitting, what should I do to solve this problem?</p>",
      "rawMarkdown": "Hello kaggler. i tuned hyperparameter tuning using bayesian optimization like link(https://www.kaggle.com/chocozzz/bayesian-optimization-for-lightgbm/code?scriptVersionId=5962809)\n\nBut the results is mystery. \n\n## Result\n- https://www.kaggle.com/prashantkikani/rstudio-lgb-single-model-lb1-6607 is \nLB: 15320 ==&gt; 15336\nCV: 1.60552376 ==&gt; 1.60596768\n\n- https://www.kaggle.com/youhanlee/stratified-sampling-for-regression-lb-1-6595 is \nLB : 15232. ==&gt; 15324\nCV: 1.62005868 ==&gt; 1.60345295\n\nI understand that the first cv and lb are both off the wrong optimization than the existing tuning value. But the second time cv got better, lb dropped. Is this an overfitting? If I am overfitting, what should I do to solve this problem?\n"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "393207": "Hello kaggler. i tuned hyperparameter tuning using bayesian optimization like link(https://www.kaggle.com/chocozzz/bayesian-optimization-for-lightgbm/code?scriptVersionId=5962809)\n\nBut the results is mystery. \n\n## Result\n- https://www.kaggle.com/prashantkikani/rstudio-lgb-single-model-lb1-6607 is \nLB: 15320 ==&gt; 15336\nCV: 1.60552376 ==&gt; 1.60596768\n\n- https://www.kaggle.com/youhanlee/stratified-sampling-for-regression-lb-1-6595 is \nLB : 15232. ==&gt; 15324\nCV: 1.62005868 ==&gt; 1.60345295\n\nI understand that the first cv and lb are both off the wrong optimization than the existing tuning value. But the second time cv got better, lb dropped. Is this an overfitting? If I am overfitting, what should I do to solve this problem?\n"
  }
}