Is this now codin competition or do I need to watch myself what they're doing?
", "messages": [ { "id": "116939", "postDate": "04/26/2016 16:02:06", "content": "Is this now codin competition or do I need to watch myself what they're doing?
", "rawMarkdown": "Is this now codin competition or do I need to watch myself what they're doing?", "votes": null }, { "id": "116971", "postDate": "04/26/2016 18:52:32", "content": "For this competition, like most Kaggle competitions, you need to deliver:
\n\n1) Prediction data in a CSV file
\n\n2) Code that produces that prediction data from the training data and test data
\n\nYou only need to deliver the second item to Kaggle in order to claim a prize. Most people do not. But if you want a chance to win, you should bear in mind that you should be able to supply code if your results are good enough.
\n\nSo strictly speaking, all Kaggle competitions are at least in part coding competitions, because you will write code - maybe based on someone else's shared script - in order to create a submission.
\n\nI'm guessing your "watch myself what they're doing?" means you think you need to look at the test images and categorise them.
\n\nThat is not the right idea.
\n\nIt might actually be possible to do this, and get a good result in the competition. But it would be a complete waste of your time. The correct categories are already known, the sponsor does not need you to supply new labels. There is no prize available for best person at guessing the correct categories by looking at the images.
\n\nThere are no competitions like that on Kaggle. If you are looking for work like that, then perhaps a better place would be https://www.mturk.com/mturk/
", "rawMarkdown": "For this competition, like most Kaggle competitions, you need to deliver:\r\n\r\n1) Prediction data in a CSV file\r\n\r\n2) Code that produces that prediction data from the training data and test data\r\n\r\nYou only need to deliver the second item to Kaggle in order to claim a prize. Most people do not. But if you want a chance to win, you should bear in mind that you should be *able* to supply code if your results are good enough.\r\n\r\nSo strictly speaking, all Kaggle competitions are at least in part coding competitions, because you will write code - maybe based on someone else's shared script - in order to create a submission.\r\n\r\n----\r\n\r\nI'm guessing your \"watch myself what they're doing?\" means you think you need to look at the test images and categorise them.\r\n\r\nThat is not the right idea.\r\n\r\nIt might actually be possible to do this, and get a good result in the competition. But it would be a complete waste of your time. The correct categories are already known, the sponsor does not need you to supply new labels. There is no prize available for best person at guessing the correct categories by looking at the images.\r\n\r\nThere are *no* competitions like that on Kaggle. If you are looking for work like that, then perhaps a better place would be https://www.mturk.com/mturk/", "votes": null } ], "comments": [ { "id": 116971, "author_name": "slobo777", "author_url": "", "post_date": "04/26/2016 18:52:32", "content": "For this competition, like most Kaggle competitions, you need to deliver:
\n\n1) Prediction data in a CSV file
\n\n2) Code that produces that prediction data from the training data and test data
\n\nYou only need to deliver the second item to Kaggle in order to claim a prize. Most people do not. But if you want a chance to win, you should bear in mind that you should be able to supply code if your results are good enough.
\n\nSo strictly speaking, all Kaggle competitions are at least in part coding competitions, because you will write code - maybe based on someone else's shared script - in order to create a submission.
\n\nI'm guessing your "watch myself what they're doing?" means you think you need to look at the test images and categorise them.
\n\nThat is not the right idea.
\n\nIt might actually be possible to do this, and get a good result in the competition. But it would be a complete waste of your time. The correct categories are already known, the sponsor does not need you to supply new labels. There is no prize available for best person at guessing the correct categories by looking at the images.
\n\nThere are no competitions like that on Kaggle. If you are looking for work like that, then perhaps a better place would be https://www.mturk.com/mturk/
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "116939": "Is this now codin competition or do I need to watch myself what they're doing?", "116971": "For this competition, like most Kaggle competitions, you need to deliver:\r\n\r\n1) Prediction data in a CSV file\r\n\r\n2) Code that produces that prediction data from the training data and test data\r\n\r\nYou only need to deliver the second item to Kaggle in order to claim a prize. Most people do not. But if you want a chance to win, you should bear in mind that you should be *able* to supply code if your results are good enough.\r\n\r\nSo strictly speaking, all Kaggle competitions are at least in part coding competitions, because you will write code - maybe based on someone else's shared script - in order to create a submission.\r\n\r\n----\r\n\r\nI'm guessing your \"watch myself what they're doing?\" means you think you need to look at the test images and categorise them.\r\n\r\nThat is not the right idea.\r\n\r\nIt might actually be possible to do this, and get a good result in the competition. But it would be a complete waste of your time. The correct categories are already known, the sponsor does not need you to supply new labels. There is no prize available for best person at guessing the correct categories by looking at the images.\r\n\r\nThere are *no* competitions like that on Kaggle. If you are looking for work like that, then perhaps a better place would be https://www.mturk.com/mturk/" }, "source": "meta" }