{
  "id": 21023,
  "title": "PCA+SVM, LB score of 1.75",
  "url": "/competitions/state-farm-distracted-driver-detection/discussion/21023",
  "author_name": "vivek yadav",
  "post_date": "2016-05-17T18:06:55.157000",
  "votes": 0,
  "comment_count": 0,
  "views": 481,
  "content": "<p>Here is my code for PCA+SVM. This got me lb score of 1.75. I will update the forum as i work on this more. For some reason, the script on kaggle is not running right. The codes below took about 20 minutes to pre-process data, and 15 minutes to make model and prediction. I am using a 2011 macbook pro with 8GB ram and NVDIA gpu processing. </p>\n\n<p>Preprocess data: Converts image to gray scale, resizes them to 48X64, and makes them into vectors of 3072X1. \n<a href=\"https://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_Preprocess_traintest.ipynb\">https://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_Preprocess_traintest.ipynb</a></p>\n\n<p>PCA + SVM: Converts image to gray scale, resizes them to 48X64, and makes them into vectors of 3072X1. \n<a href=\"https://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_ImPrcess_pkl_SVC.ipynb\">https://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_ImPrcess_pkl_SVC.ipynb</a></p>",
  "messages": [
    {
      "id": 120353,
      "postDate": "2016-05-17T18:06:55.157Z",
      "content": "<p>Here is my code for PCA+SVM. This got me lb score of 1.75. I will update the forum as i work on this more. For some reason, the script on kaggle is not running right. The codes below took about 20 minutes to pre-process data, and 15 minutes to make model and prediction. I am using a 2011 macbook pro with 8GB ram and NVDIA gpu processing. </p>\n\n<p>Preprocess data: Converts image to gray scale, resizes them to 48X64, and makes them into vectors of 3072X1. \n<a href=\"https://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_Preprocess_traintest.ipynb\">https://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_Preprocess_traintest.ipynb</a></p>\n\n<p>PCA + SVM: Converts image to gray scale, resizes them to 48X64, and makes them into vectors of 3072X1. \n<a href=\"https://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_ImPrcess_pkl_SVC.ipynb\">https://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_ImPrcess_pkl_SVC.ipynb</a></p>",
      "rawMarkdown": "Here is my code for PCA+SVM. This got me lb score of 1.75. I will update the forum as i work on this more. For some reason, the script on kaggle is not running right. The codes below took about 20 minutes to pre-process data, and 15 minutes to make model and prediction. I am using a 2011 macbook pro with 8GB ram and NVDIA gpu processing. \r\n\r\nPreprocess data: Converts image to gray scale, resizes them to 48X64, and makes them into vectors of 3072X1. \r\nhttps://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_Preprocess_traintest.ipynb\r\n\r\nPCA + SVM: Converts image to gray scale, resizes them to 48X64, and makes them into vectors of 3072X1. \r\nhttps://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_ImPrcess_pkl_SVC.ipynb\r\n\r\n"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "120353": "Here is my code for PCA+SVM. This got me lb score of 1.75. I will update the forum as i work on this more. For some reason, the script on kaggle is not running right. The codes below took about 20 minutes to pre-process data, and 15 minutes to make model and prediction. I am using a 2011 macbook pro with 8GB ram and NVDIA gpu processing. \r\n\r\nPreprocess data: Converts image to gray scale, resizes them to 48X64, and makes them into vectors of 3072X1. \r\nhttps://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_Preprocess_traintest.ipynb\r\n\r\nPCA + SVM: Converts image to gray scale, resizes them to 48X64, and makes them into vectors of 3072X1. \r\nhttps://nbviewer.jupyter.org/github/vxy10/Kaggle_GH_Vivek/blob/master/StateFarm/main_SFrm_ImPrcess_pkl_SVC.ipynb\r\n\r\n"
  }
}