{
  "id": 35259,
  "title": "Feedback about sklearn.neural_network experiment",
  "url": "/competitions/digit-recognizer/discussion/35259",
  "author_name": "Eric Vos",
  "post_date": "2017-06-25T12:44:18.275000",
  "votes": 1,
  "comment_count": 1,
  "views": null,
  "content": "<p>Hello,</p>\n\n<p>I coded quickly a solution based on sklearn.neural_network-MLPClassifier with following parameters:</p>\n\n<pre><code>clf = MLPClassifier(hidden_layer_sizes = [200, 200], alpha = 5.0,random_state = 0, solver='lbfgs').fit(X_train, y_train)\n</code></pre>\n\n<p>Code is published <a href=\"https://github.com/experimentboy/MNIST_experiment\">here</a>\nResult of the code is 0.93871.</p>\n\n<p>Any comments about potential improvement is welcome.\nIn advance, Thank you\nBest regards, Eric</p>",
  "messages": [
    {
      "id": 195871,
      "postDate": "2017-06-25T12:44:18.277Z",
      "content": "<p>Hello,</p>\n\n<p>I coded quickly a solution based on sklearn.neural_network-MLPClassifier with following parameters:</p>\n\n<pre><code>clf = MLPClassifier(hidden_layer_sizes = [200, 200], alpha = 5.0,random_state = 0, solver='lbfgs').fit(X_train, y_train)\n</code></pre>\n\n<p>Code is published <a href=\"https://github.com/experimentboy/MNIST_experiment\">here</a>\nResult of the code is 0.93871.</p>\n\n<p>Any comments about potential improvement is welcome.\nIn advance, Thank you\nBest regards, Eric</p>",
      "rawMarkdown": "Hello,\n\nI coded quickly a solution based on sklearn.neural_network-MLPClassifier with following parameters:\n\n    clf = MLPClassifier(hidden_layer_sizes = [200, 200], alpha = 5.0,random_state = 0, solver='lbfgs').fit(X_train, y_train)\n\nCode is published [here][1]\nResult of the code is 0.93871.\n\nAny comments about potential improvement is welcome.\nIn advance, Thank you\nBest regards, Eric\n\n\n  [1]: https://github.com/experimentboy/MNIST_experiment",
      "votes": 1
    },
    {
      "id": 196605,
      "postDate": "2017-06-27T19:25:28.657Z",
      "content": "<p>Well done. I just miss the rationale that led you selecting those optimal parameters :-) You owe me an explanation around a coffee!</p>",
      "rawMarkdown": "Well done. I just miss the rationale that led you selecting those optimal parameters :-) You owe me an explanation around a coffee!"
    }
  ],
  "comments": [
    {
      "id": 196605,
      "author_name": "Xavier Mayeur",
      "author_url": "",
      "post_date": "2017-06-27T19:25:28.657000",
      "content": "<p>Well done. I just miss the rationale that led you selecting those optimal parameters :-) You owe me an explanation around a coffee!</p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "195871": "Hello,\n\nI coded quickly a solution based on sklearn.neural_network-MLPClassifier with following parameters:\n\n    clf = MLPClassifier(hidden_layer_sizes = [200, 200], alpha = 5.0,random_state = 0, solver='lbfgs').fit(X_train, y_train)\n\nCode is published [here][1]\nResult of the code is 0.93871.\n\nAny comments about potential improvement is welcome.\nIn advance, Thank you\nBest regards, Eric\n\n\n  [1]: https://github.com/experimentboy/MNIST_experiment",
    "196605": "Well done. I just miss the rationale that led you selecting those optimal parameters :-) You owe me an explanation around a coffee!"
  }
}