{
  "id": 556637,
  "title": "Ways to improve the model",
  "url": "/competitions/jane-street-real-time-market-data-forecasting/discussion/556637",
  "author_name": "",
  "post_date": "2025-01-14T11:37:36.483506200Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p><strong>I want to know more about the ways anybody used improve the model performance</strong><br>\nMy ways are as follows:</p>\n<p>Throughout the competition, I have thought of the following ways to improve the models' performance:</p>\n<ol>\n<li>Ensemble existing Notebooks.</li>\n<li>Adjust model training process: <br>\n(2.1)Fine Tuning hyper-parameters; <br>\n(2.2)Download the data and use the whole dataset to train XGB, LightGBM, Catboost models (on my own GPU);<br>\n(2.3)Try different models(Random Forest, Linear Regression…)</li>\n<li>In the final predict function, remove some highly correlated features</li>\n</ol>\n<p>I am looking forward to your sharing about ways to improve the models' performance!</p>",
  "messages": [
    {
      "id": "3096429",
      "postDate": "01/14/2025 11:37:36",
      "content": "<p><strong>I want to know more about the ways anybody used improve the model performance</strong><br>\nMy ways are as follows:</p>\n<p>Throughout the competition, I have thought of the following ways to improve the models' performance:</p>\n<ol>\n<li>Ensemble existing Notebooks.</li>\n<li>Adjust model training process: <br>\n(2.1)Fine Tuning hyper-parameters; <br>\n(2.2)Download the data and use the whole dataset to train XGB, LightGBM, Catboost models (on my own GPU);<br>\n(2.3)Try different models(Random Forest, Linear Regression…)</li>\n<li>In the final predict function, remove some highly correlated features</li>\n</ol>\n<p>I am looking forward to your sharing about ways to improve the models' performance!</p>",
      "rawMarkdown": "**I want to know more about the ways anybody used improve the model performance**\nMy ways are as follows:\n\nThroughout the competition, I have thought of the following ways to improve the models' performance:\n1. Ensemble existing Notebooks.\n2. Adjust model training process: \n  (2.1)Fine Tuning hyper-parameters; \n  (2.2)Download the data and use the whole dataset to train XGB, LightGBM, Catboost models (on my own GPU);\n  (2.3)Try different models(Random Forest, Linear Regression...)\n3. In the final predict function, remove some highly correlated features\n\nI am looking forward to your sharing about ways to improve the models' performance!",
      "votes": null
    },
    {
      "id": "3099055",
      "postDate": "01/17/2025 08:45:55",
      "content": "<p>I am looking forward to your sharing about ways to improve the models' performance!</p>",
      "rawMarkdown": "I am looking forward to your sharing about ways to improve the models' performance!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3099055,
      "author_name": "xuchenjames",
      "author_url": "",
      "post_date": "01/17/2025 08:45:55",
      "content": "<p>I am looking forward to your sharing about ways to improve the models' performance!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3096429": "**I want to know more about the ways anybody used improve the model performance**\nMy ways are as follows:\n\nThroughout the competition, I have thought of the following ways to improve the models' performance:\n1. Ensemble existing Notebooks.\n2. Adjust model training process: \n  (2.1)Fine Tuning hyper-parameters; \n  (2.2)Download the data and use the whole dataset to train XGB, LightGBM, Catboost models (on my own GPU);\n  (2.3)Try different models(Random Forest, Linear Regression...)\n3. In the final predict function, remove some highly correlated features\n\nI am looking forward to your sharing about ways to improve the models' performance!",
    "3099055": "I am looking forward to your sharing about ways to improve the models' performance!"
  },
  "source": "meta"
}