I am new to Machine and Deep learning, and in class with been asked to solve the competition using deep learning (we use use a neural network), however i find it rather difficult.
\n\nAny suggestions, some notebooks start up code, or maybe how to proceed?\nI am not concerned with the accuracy rather than the implementation, i tried a simple neural network for regression, however i get Nan in the loss function.
\n\nAny guidance, instructions or starter code (preferably ) is much appreciated. Thanks
", "messages": [ { "id": "702866", "postDate": "12/25/2019 09:27:48", "content": "I am new to Machine and Deep learning, and in class with been asked to solve the competition using deep learning (we use use a neural network), however i find it rather difficult.
\n\nAny suggestions, some notebooks start up code, or maybe how to proceed?\nI am not concerned with the accuracy rather than the implementation, i tried a simple neural network for regression, however i get Nan in the loss function.
\n\nAny guidance, instructions or starter code (preferably ) is much appreciated. Thanks
", "rawMarkdown": "I am new to Machine and Deep learning, and in class with been asked to solve the competition using deep learning (we use use a neural network), however i find it rather difficult.\n\nAny suggestions, some notebooks start up code, or maybe how to proceed?\nI am not concerned with the accuracy rather than the implementation, i tried a simple neural network for regression, however i get Nan in the loss function.\n\n\nAny guidance, instructions or starter code (preferably ) is much appreciated. Thanks", "votes": null }, { "id": "702917", "postDate": "12/25/2019 10:58:56", "content": "You have to predict 2 years with 1 years of data. So it's a bit different to do it with LSTM for example. I would suggest lightgbm or xgboost for this competition.\nThis kernel should help you if you are looking only for the implementation and not in the performance of the model.
", "rawMarkdown": "You have to predict 2 years with 1 years of data. So it's a bit different to do it with LSTM for example. I would suggest lightgbm or xgboost for this competition.\n[This kernel](https://www.kaggle.com/drcapa/ashrae-datagenerator-lstm) should help you if you are looking only for the implementation and not in the performance of the model.", "votes": null }, { "id": "703241", "postDate": "12/25/2019 21:15:48", "content": "You could look at https://www.kaggle.com/isaienkov/keras-nn-with-embeddings-for-cat-features-1-15 to start with. \nTo improve it you need to do some data cleanup.
", "rawMarkdown": "You could look at https://www.kaggle.com/isaienkov/keras-nn-with-embeddings-for-cat-features-1-15 to start with. \nTo improve it you need to do some data cleanup.", "votes": null }, { "id": "703784", "postDate": "12/26/2019 16:08:26", "content": "Simple and useful solution ! https://www.kaggle.com/c/ashrae-energy-prediction/discussion/123288
\n\nKernel by @wittmannf : https://www.kaggle.com/wittmannf/0-939-lb-public-blend-leak-valid-ridgecv/
", "rawMarkdown": "Simple and useful solution ! https://www.kaggle.com/c/ashrae-energy-prediction/discussion/123288\n\nKernel by @wittmannf : https://www.kaggle.com/wittmannf/0-939-lb-public-blend-leak-valid-ridgecv/", "votes": null } ], "comments": [ { "id": 702917, "author_name": "mimoudata", "author_url": "", "post_date": "12/25/2019 10:58:56", "content": "You have to predict 2 years with 1 years of data. So it's a bit different to do it with LSTM for example. I would suggest lightgbm or xgboost for this competition.\nThis kernel should help you if you are looking only for the implementation and not in the performance of the model.
", "votes": null, "replies": [] }, { "id": 703241, "author_name": "helgith", "author_url": "", "post_date": "12/25/2019 21:15:48", "content": "You could look at https://www.kaggle.com/isaienkov/keras-nn-with-embeddings-for-cat-features-1-15 to start with. \nTo improve it you need to do some data cleanup.
", "votes": null, "replies": [] }, { "id": 703784, "author_name": "caesarlupum", "author_url": "", "post_date": "12/26/2019 16:08:26", "content": "Simple and useful solution ! https://www.kaggle.com/c/ashrae-energy-prediction/discussion/123288
\n\nKernel by @wittmannf : https://www.kaggle.com/wittmannf/0-939-lb-public-blend-leak-valid-ridgecv/
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "702866": "I am new to Machine and Deep learning, and in class with been asked to solve the competition using deep learning (we use use a neural network), however i find it rather difficult.\n\nAny suggestions, some notebooks start up code, or maybe how to proceed?\nI am not concerned with the accuracy rather than the implementation, i tried a simple neural network for regression, however i get Nan in the loss function.\n\n\nAny guidance, instructions or starter code (preferably ) is much appreciated. Thanks", "702917": "You have to predict 2 years with 1 years of data. So it's a bit different to do it with LSTM for example. I would suggest lightgbm or xgboost for this competition.\n[This kernel](https://www.kaggle.com/drcapa/ashrae-datagenerator-lstm) should help you if you are looking only for the implementation and not in the performance of the model.", "703241": "You could look at https://www.kaggle.com/isaienkov/keras-nn-with-embeddings-for-cat-features-1-15 to start with. \nTo improve it you need to do some data cleanup.", "703784": "Simple and useful solution ! https://www.kaggle.com/c/ashrae-energy-prediction/discussion/123288\n\nKernel by @wittmannf : https://www.kaggle.com/wittmannf/0-939-lb-public-blend-leak-valid-ridgecv/" }, "source": "meta" }