{ "id": 77122, "title": "What should a newbie do?", "url": "/competitions/quora-insincere-questions-classification/discussion/77122", "author_name": "Vladyslav Bilotserkovskyi", "post_date": "2019-01-09T19:58:14.063000", "votes": 10, "comment_count": 18, "views": 0, "content": "

Hi everybody, I'm a newbie in Machine Learning and in 'Kaggling'. And I want to ask experienced kegglers and data scientists to give some pieces of advice on how a newbie can improve his skills?\nI know that it is a lot of posts throw the internet about it, but it will be cool if some guys with real experience share the list of technologies they commonly use, tasks that they usually work on. \nIt will be perfect if someone will share links on useful things, because of some kind of specific papers usually hard to understand, official documentation can't give all answers and so on.\nAlso, it will be cool to get some advice about the best ways of 'kaggling')) Everyone wants to take a medal but I read, that kaggle, first of all, it's a platform for learning and sharing (I want to say very big 'Thank You' for all guys who make awesome kernels and answers in discussions!!)((

\n\n

P.S. sorry for my English((

", "messages": [ { "id": 453492, "postDate": "2019-01-10T09:19:55.063Z", "content": "

I was pretty newbie until last summer. I have started my competitions journey via Zillow House price. I teamed up and learn the subtle arts of;

\n\n

Trust Your Cross Validation. Always. Period \n- By far the most important part of competition you wont hear everyone talk because its so common and key winning trick. A good validation logic for a given problem can get you gold as well as throw positions far away from bronze medals. I was at 5th at public LB Home Credit Default Risk. My team had pretty much done CV till last point. But, the blending submission was not done on common fold leading to 35th in private LB and we did loose gold. But, I am very happy that I made it and never going to ignore such simple mistake again.

\n\n

No. We all dont have workstations with us \n- Before making to expert. I thought it be possible to earn kaggle medals only with a high tech Nvidia powered workstation. But, I was wrong (except image competitions). I still use old laptop with 128 GB of space with 4/8 GB of RAM. Most of the time I rely on kaggle kernels. So, advice - Dont let resources slow you. Learn optimizations and tricks which you can keep in your sleeve for long run.

\n\n

Its not how many models. But your features \n- Again, you really need to be a data scientist with 1000 models to win a competition. Not true. There are many winners and gold medallist in past competition who won just because they have engineering awesome features spend on domain knowledge hunting and grabbed a gold medal. In one competition @Giba's single LGBM is at 1st private LB position too. How about that ?

\n\n

Don't compete to win. Compete for sportivness and learn \n- Many of kagglers aren't behind medals. That are just rewards kaggle recognises for a user. For many kaggle is a hobby. A late night fun and for others its a project. In the end .00000X accuracies doesn't matter, what matters is what you learnt from this competition. Take away the good. Don't repeat mistakes of past. And when you attend any new competition again, you are a Kaggler!

\n\n

Attaching a scene from Avengers. I felt like that at start

\n\n

\"\"

", "rawMarkdown": "I was pretty newbie until last summer. I have started my competitions journey via Zillow House price. I teamed up and learn the subtle arts of;\n\n**Trust Your Cross Validation. Always. Period** \n- By far the most important part of competition you wont hear everyone talk because its so common and key winning trick. A good validation logic for a given problem can get you gold as well as throw positions far away from bronze medals. I was at 5th at public LB Home Credit Default Risk. My team had pretty much done CV till last point. But, the blending submission was not done on common fold leading to 35th in private LB and we did loose gold. But, I am very happy that I made it and never going to ignore such simple mistake again. \n\n**No. We all dont have workstations with us** \n- Before making to expert. I thought it be possible to earn kaggle medals only with a high tech Nvidia powered workstation. But, I was wrong (except image competitions). I still use old laptop with 128 GB of space with 4/8 GB of RAM. Most of the time I rely on kaggle kernels. So, advice - Dont let resources slow you. Learn optimizations and tricks which you can keep in your sleeve for long run. \n\n**Its not how many models. But your features** \n- Again, you really need to be a data scientist with 1000 models to win a competition. Not true. There are many winners and gold medallist in past competition who won just because they have engineering awesome features spend on domain knowledge hunting and grabbed a gold medal. In one competition @Giba's single LGBM is at 1st private LB position too. How about that ? \n\n**Don't compete to win. Compete for sportivness and learn** \n- Many of kagglers aren't behind medals. That are just rewards kaggle recognises for a user. For many kaggle is a hobby. A late night fun and for others its a project. In the end .00000X accuracies doesn't matter, what matters is what you learnt from this competition. Take away the good. Don't repeat mistakes of past. And when you attend any new competition again, you are a Kaggler! \n\n*Attaching a scene from Avengers. I felt like that at start*\n\n![](https://i.redd.it/ay5cfurj0f301.jpg)", "votes": 25, "replies": [ { "id": 453525, "postDate": "2019-01-10T10:09:22.987Z", "rawMarkdown": "", "isDeleted": true }, { "id": 453576, "postDate": "2019-01-10T12:07:57.307Z", "content": "

Thank You for the very cool reply!

", "rawMarkdown": "Thank You for the very cool reply! " }, { "id": 453594, "postDate": "2019-01-10T12:58:59.143Z", "content": "

I followed you. Looking forward to see your profile in expert tier theme. Good luck @psyfaker!

", "rawMarkdown": "I followed you. Looking forward to see your profile in expert tier theme. Good luck @psyfaker! ", "votes": 2 }, { "id": 453606, "postDate": "2019-01-10T13:24:39.877Z", "content": "

Thank You very much!! Good Luck with a Master rank!!!)))

", "rawMarkdown": "Thank You very much!! Good Luck with a Master rank!!!)))" } ] }, { "id": 453174, "postDate": "2019-01-09T19:58:14.063Z", "content": "

Hi everybody, I'm a newbie in Machine Learning and in 'Kaggling'. And I want to ask experienced kegglers and data scientists to give some pieces of advice on how a newbie can improve his skills?\nI know that it is a lot of posts throw the internet about it, but it will be cool if some guys with real experience share the list of technologies they commonly use, tasks that they usually work on. \nIt will be perfect if someone will share links on useful things, because of some kind of specific papers usually hard to understand, official documentation can't give all answers and so on.\nAlso, it will be cool to get some advice about the best ways of 'kaggling')) Everyone wants to take a medal but I read, that kaggle, first of all, it's a platform for learning and sharing (I want to say very big 'Thank You' for all guys who make awesome kernels and answers in discussions!!)((

\n\n

P.S. sorry for my English((

", "rawMarkdown": "Hi everybody, I'm a newbie in Machine Learning and in 'Kaggling'. And I want to ask experienced kegglers and data scientists to give some pieces of advice on how a newbie can improve his skills?\nI know that it is a lot of posts throw the internet about it, but it will be cool if some guys with real experience share the list of technologies they commonly use, tasks that they usually work on. \nIt will be perfect if someone will share links on useful things, because of some kind of specific papers usually hard to understand, official documentation can't give all answers and so on.\nAlso, it will be cool to get some advice about the best ways of 'kaggling')) Everyone wants to take a medal but I read, that kaggle, first of all, it's a platform for learning and sharing (I want to say very big 'Thank You' for all guys who make awesome kernels and answers in discussions!!)((\n\nP.S. sorry for my English((", "votes": 9 }, { "id": 453720, "postDate": "2019-01-10T16:54:12.393Z", "content": "

I'd recommend to join our upcoming session of http://mlcourse.ai - open ML course by OpenDataScience. This time, we'll have 4 competitions (Kaggle Inclass and other platforms) during the course. We'll provide benchmarks and step-by-step guides how to beat them. The previous announcement is here on Kaggle https://www.kaggle.com/general/68205 though it's a bit different, the new session is much more competition-focused.\nps. considering that you can read Russian: this article is exactly about kaggle best practices https://habr.com/company/ods/blog/426227/

", "rawMarkdown": "I'd recommend to join our upcoming session of http://mlcourse.ai - open ML course by OpenDataScience. This time, we'll have 4 competitions (Kaggle Inclass and other platforms) during the course. We'll provide benchmarks and step-by-step guides how to beat them. The previous announcement is here on Kaggle https://www.kaggle.com/general/68205 though it's a bit different, the new session is much more competition-focused.\nps. considering that you can read Russian: this article is exactly about kaggle best practices https://habr.com/company/ods/blog/426227/", "votes": 5 }, { "id": 455547, "postDate": "2019-01-14T06:26:30.443Z", "content": "

I am also a newbie in this. In my course curriculum, i studied data analytics in my postgraduate program in industrial engineering and management. From there my journey started. i got more interested in this topic, then my professor told me to start learning R or python, It will be helpful for doing analysis. After the end of this course my professor suggested me a course on machine learning by Andrew Ng on Course. I have completed this course and now am learning deep learning by the same professor on coursera. From kaggle also i am learning. I am participating in the competition and improving day by day. You also follow the same path.

", "rawMarkdown": "I am also a newbie in this. In my course curriculum, i studied data analytics in my postgraduate program in industrial engineering and management. From there my journey started. i got more interested in this topic, then my professor told me to start learning R or python, It will be helpful for doing analysis. After the end of this course my professor suggested me a course on machine learning by Andrew Ng on Course. I have completed this course and now am learning deep learning by the same professor on coursera. From kaggle also i am learning. I am participating in the competition and improving day by day. You also follow the same path.", "votes": 1, "replies": [ { "id": 593154, "postDate": "2019-08-06T08:21:05.653Z", "content": "

I cant find the ML course by Andrew that you are talking about. Can you please upload the link? It would be really useful for me, as I am also starting. I have tried the Intro to ML by Kaggle, but I feel as if I am not learning anything, as the course shows you a way to do something, and in the exercise it changes it. that may not be the case, but that is how I feel. If you have any suggestions on courses I can follow to get started in ML, I would be really grateful...👍

", "rawMarkdown": "I cant find the ML course by Andrew that you are talking about. Can you please upload the link? It would be really useful for me, as I am also starting. I have tried the Intro to ML by Kaggle, but I feel as if I am not learning anything, as the course shows you a way to do something, and in the exercise it changes it. that may not be the case, but that is how I feel. If you have any suggestions on courses I can follow to get started in ML, I would be really grateful...👍 " }, { "id": 593199, "postDate": "2019-08-06T09:47:52.867Z", "content": "

Here it is.

\n\n

Though I'd recommend a modern practical course mlcourse.ai. Andrew Ng's course is very nice, but it's outdated.

", "rawMarkdown": "[Here it is](https://www.coursera.org/learn/machine-learning).\n\nThough I'd recommend a modern practical course [mlcourse.ai](https://mlcourse.ai). Andrew Ng's course is very nice, but it's outdated. " } ] }, { "id": 453338, "postDate": "2019-01-10T02:58:55.480Z", "content": "

Hi Vladyslav, since almost everyone has a different definition for newbie. Could you please share your background on machine learning ? (what have you learned up until now, so that some experts can understand you better, and they can give you appropriate directions)

", "rawMarkdown": "Hi Vladyslav, since almost everyone has a different definition for newbie. Could you please share your background on machine learning ? (what have you learned up until now, so that some experts can understand you better, and they can give you appropriate directions)", "votes": 1, "replies": [ { "id": 453393, "postDate": "2019-01-10T05:58:00.447Z", "content": "

I completed deep learning ai courses by Andrew Ng. \nhttps://www.coursera.org/specializations/deep-learning\nAlso I Know some basic math and statistic but not advance things. I'm trying to improve my skills with kaggle. Preferably I work on NLP and image recognition tasks so I'm absolutely noob in basic classifications tasks.

", "rawMarkdown": "I completed deep learning ai courses by Andrew Ng. \nhttps://www.coursera.org/specializations/deep-learning\nAlso I Know some basic math and statistic but not advance things. I'm trying to improve my skills with kaggle. Preferably I work on NLP and image recognition tasks so I'm absolutely noob in basic classifications tasks.", "votes": 1 }, { "id": 453395, "postDate": "2019-01-10T06:03:49.760Z", "content": "

I created this post not for some concrete advice for me, but for experience developers could share information about most popular tasks and frameworks that they use what kind of frameworks they mostly use when they have some problems with an implementation where they look for information. Because data science it's a very big area to research, so should we start learning it step by step like all post on hard describe or maybe we can start only from the most popular areas?

", "rawMarkdown": "I created this post not for some concrete advice for me, but for experience developers could share information about most popular tasks and frameworks that they use what kind of frameworks they mostly use when they have some problems with an implementation where they look for information. Because data science it's a very big area to research, so should we start learning it step by step like all post on hard describe or maybe we can start only from the most popular areas?", "votes": 1 } ] }, { "id": 453197, "postDate": "2019-01-09T20:47:23.637Z", "content": "

I'm also a newbie but this is a one of the great courses of machine learning and deep learning :\nhttps://course.fast.ai/ml.html

", "rawMarkdown": "I'm also a newbie but this is a one of the great courses of machine learning and deep learning :\nhttps://course.fast.ai/ml.html", "votes": 1, "replies": [ { "id": 453390, "postDate": "2019-01-10T05:53:01.663Z", "content": "

Thanks for sharing, fastai it's a very cool tool. As I know, it's preferably used for image recognition task or I'm wrong?

", "rawMarkdown": "Thanks for sharing, fastai it's a very cool tool. As I know, it's preferably used for image recognition task or I'm wrong?" }, { "id": 453405, "postDate": "2019-01-10T06:32:52.817Z", "content": "

Hi again Vladyslav, fast.ai also teaches great NLP tools such as ULMFit which having state of the art performances of many NLP classification tasks.

", "rawMarkdown": "Hi again Vladyslav, fast.ai also teaches great NLP tools such as ULMFit which having state of the art performances of many NLP classification tasks.", "votes": 1 }, { "id": 453408, "postDate": "2019-01-10T06:40:45.723Z", "content": "

O, cool) So question is which of frameworks is better to learn? I want to learn pytorch because it's really fast and after this fast ai because it has a lot of cool things in it, is it a good decision?\nhttps://www.edx.org/course/deep-learning-with-python-and-pytorch

", "rawMarkdown": "O, cool) So question is which of frameworks is better to learn? I want to learn pytorch because it's really fast and after this fast ai because it has a lot of cool things in it, is it a good decision?\nhttps://www.edx.org/course/deep-learning-with-python-and-pytorch" }, { "id": 453518, "postDate": "2019-01-10T09:57:06.763Z", "content": "

fast.ai also mainly uses pytorch, so you cannot be wrong either way ;)

", "rawMarkdown": "fast.ai also mainly uses pytorch, so you cannot be wrong either way ;)", "votes": 2 }, { "id": 507235, "postDate": "2019-04-04T13:18:44.360Z", "content": "

thank u

", "rawMarkdown": "thank u" } ] } ], "comments": [ { "id": 453492, "author_name": "Shahebaz Mohammad", "author_url": "", "post_date": "2019-01-10T09:19:55.063000", "content": "

I was pretty newbie until last summer. I have started my competitions journey via Zillow House price. I teamed up and learn the subtle arts of;

\n\n

Trust Your Cross Validation. Always. Period \n- By far the most important part of competition you wont hear everyone talk because its so common and key winning trick. A good validation logic for a given problem can get you gold as well as throw positions far away from bronze medals. I was at 5th at public LB Home Credit Default Risk. My team had pretty much done CV till last point. But, the blending submission was not done on common fold leading to 35th in private LB and we did loose gold. But, I am very happy that I made it and never going to ignore such simple mistake again.

\n\n

No. We all dont have workstations with us \n- Before making to expert. I thought it be possible to earn kaggle medals only with a high tech Nvidia powered workstation. But, I was wrong (except image competitions). I still use old laptop with 128 GB of space with 4/8 GB of RAM. Most of the time I rely on kaggle kernels. So, advice - Dont let resources slow you. Learn optimizations and tricks which you can keep in your sleeve for long run.

\n\n

Its not how many models. But your features \n- Again, you really need to be a data scientist with 1000 models to win a competition. Not true. There are many winners and gold medallist in past competition who won just because they have engineering awesome features spend on domain knowledge hunting and grabbed a gold medal. In one competition @Giba's single LGBM is at 1st private LB position too. How about that ?

\n\n

Don't compete to win. Compete for sportivness and learn \n- Many of kagglers aren't behind medals. That are just rewards kaggle recognises for a user. For many kaggle is a hobby. A late night fun and for others its a project. In the end .00000X accuracies doesn't matter, what matters is what you learnt from this competition. Take away the good. Don't repeat mistakes of past. And when you attend any new competition again, you are a Kaggler!

\n\n

Attaching a scene from Avengers. I felt like that at start

\n\n

\"\"

", "votes": 25, "replies": [ { "id": 453525, "author_name": "", "author_url": "", "post_date": "2019-01-10T10:09:22.987000", "content": "", "votes": 0, "replies": [] }, { "id": 453576, "author_name": "Vladyslav Bilotserkovskyi", "author_url": "", "post_date": "2019-01-10T12:07:57.307000", "content": "

Thank You for the very cool reply!

", "votes": 0, "replies": [] }, { "id": 453594, "author_name": "Shahebaz Mohammad", "author_url": "", "post_date": "2019-01-10T12:58:59.143000", "content": "

I followed you. Looking forward to see your profile in expert tier theme. Good luck @psyfaker!

", "votes": 2, "replies": [] }, { "id": 453606, "author_name": "Vladyslav Bilotserkovskyi", "author_url": "", "post_date": "2019-01-10T13:24:39.877000", "content": "

Thank You very much!! Good Luck with a Master rank!!!)))

", "votes": 0, "replies": [] } ] }, { "id": 453720, "author_name": "Yury Kashnitsky", "author_url": "", "post_date": "2019-01-10T16:54:12.393000", "content": "

I'd recommend to join our upcoming session of http://mlcourse.ai - open ML course by OpenDataScience. This time, we'll have 4 competitions (Kaggle Inclass and other platforms) during the course. We'll provide benchmarks and step-by-step guides how to beat them. The previous announcement is here on Kaggle https://www.kaggle.com/general/68205 though it's a bit different, the new session is much more competition-focused.\nps. considering that you can read Russian: this article is exactly about kaggle best practices https://habr.com/company/ods/blog/426227/

", "votes": 5, "replies": [] }, { "id": 455547, "author_name": "ALOK PRATAP", "author_url": "", "post_date": "2019-01-14T06:26:30.443000", "content": "

I am also a newbie in this. In my course curriculum, i studied data analytics in my postgraduate program in industrial engineering and management. From there my journey started. i got more interested in this topic, then my professor told me to start learning R or python, It will be helpful for doing analysis. After the end of this course my professor suggested me a course on machine learning by Andrew Ng on Course. I have completed this course and now am learning deep learning by the same professor on coursera. From kaggle also i am learning. I am participating in the competition and improving day by day. You also follow the same path.

", "votes": 1, "replies": [ { "id": 593154, "author_name": "Mateo Espinosa-Bravo", "author_url": "", "post_date": "2019-08-06T08:21:05.653000", "content": "

I cant find the ML course by Andrew that you are talking about. Can you please upload the link? It would be really useful for me, as I am also starting. I have tried the Intro to ML by Kaggle, but I feel as if I am not learning anything, as the course shows you a way to do something, and in the exercise it changes it. that may not be the case, but that is how I feel. If you have any suggestions on courses I can follow to get started in ML, I would be really grateful...👍

", "votes": 0, "replies": [] }, { "id": 593199, "author_name": "Yury Kashnitsky", "author_url": "", "post_date": "2019-08-06T09:47:52.867000", "content": "

Here it is.

\n\n

Though I'd recommend a modern practical course mlcourse.ai. Andrew Ng's course is very nice, but it's outdated.

", "votes": 0, "replies": [] } ] }, { "id": 453338, "author_name": "Neuron Engineer", "author_url": "", "post_date": "2019-01-10T02:58:55.480000", "content": "

Hi Vladyslav, since almost everyone has a different definition for newbie. Could you please share your background on machine learning ? (what have you learned up until now, so that some experts can understand you better, and they can give you appropriate directions)

", "votes": 1, "replies": [ { "id": 453393, "author_name": "Vladyslav Bilotserkovskyi", "author_url": "", "post_date": "2019-01-10T05:58:00.447000", "content": "

I completed deep learning ai courses by Andrew Ng. \nhttps://www.coursera.org/specializations/deep-learning\nAlso I Know some basic math and statistic but not advance things. I'm trying to improve my skills with kaggle. Preferably I work on NLP and image recognition tasks so I'm absolutely noob in basic classifications tasks.

", "votes": 1, "replies": [] }, { "id": 453395, "author_name": "Vladyslav Bilotserkovskyi", "author_url": "", "post_date": "2019-01-10T06:03:49.760000", "content": "

I created this post not for some concrete advice for me, but for experience developers could share information about most popular tasks and frameworks that they use what kind of frameworks they mostly use when they have some problems with an implementation where they look for information. Because data science it's a very big area to research, so should we start learning it step by step like all post on hard describe or maybe we can start only from the most popular areas?

", "votes": 1, "replies": [] } ] }, { "id": 453197, "author_name": "Ouassim Adnane", "author_url": "", "post_date": "2019-01-09T20:47:23.637000", "content": "

I'm also a newbie but this is a one of the great courses of machine learning and deep learning :\nhttps://course.fast.ai/ml.html

", "votes": 1, "replies": [ { "id": 453390, "author_name": "Vladyslav Bilotserkovskyi", "author_url": "", "post_date": "2019-01-10T05:53:01.663000", "content": "

Thanks for sharing, fastai it's a very cool tool. As I know, it's preferably used for image recognition task or I'm wrong?

", "votes": 0, "replies": [] }, { "id": 453405, "author_name": "Neuron Engineer", "author_url": "", "post_date": "2019-01-10T06:32:52.817000", "content": "

Hi again Vladyslav, fast.ai also teaches great NLP tools such as ULMFit which having state of the art performances of many NLP classification tasks.

", "votes": 1, "replies": [] }, { "id": 453408, "author_name": "Vladyslav Bilotserkovskyi", "author_url": "", "post_date": "2019-01-10T06:40:45.723000", "content": "

O, cool) So question is which of frameworks is better to learn? I want to learn pytorch because it's really fast and after this fast ai because it has a lot of cool things in it, is it a good decision?\nhttps://www.edx.org/course/deep-learning-with-python-and-pytorch

", "votes": 0, "replies": [] }, { "id": 453518, "author_name": "Neuron Engineer", "author_url": "", "post_date": "2019-01-10T09:57:06.763000", "content": "

fast.ai also mainly uses pytorch, so you cannot be wrong either way ;)

", "votes": 2, "replies": [] }, { "id": 507235, "author_name": "nassim", "author_url": "", "post_date": "2019-04-04T13:18:44.360000", "content": "

thank u

", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "453492": "I was pretty newbie until last summer. I have started my competitions journey via Zillow House price. I teamed up and learn the subtle arts of;\n\n**Trust Your Cross Validation. Always. Period** \n- By far the most important part of competition you wont hear everyone talk because its so common and key winning trick. A good validation logic for a given problem can get you gold as well as throw positions far away from bronze medals. I was at 5th at public LB Home Credit Default Risk. My team had pretty much done CV till last point. But, the blending submission was not done on common fold leading to 35th in private LB and we did loose gold. But, I am very happy that I made it and never going to ignore such simple mistake again. \n\n**No. We all dont have workstations with us** \n- Before making to expert. I thought it be possible to earn kaggle medals only with a high tech Nvidia powered workstation. But, I was wrong (except image competitions). I still use old laptop with 128 GB of space with 4/8 GB of RAM. Most of the time I rely on kaggle kernels. So, advice - Dont let resources slow you. Learn optimizations and tricks which you can keep in your sleeve for long run. \n\n**Its not how many models. But your features** \n- Again, you really need to be a data scientist with 1000 models to win a competition. Not true. There are many winners and gold medallist in past competition who won just because they have engineering awesome features spend on domain knowledge hunting and grabbed a gold medal. In one competition @Giba's single LGBM is at 1st private LB position too. How about that ? \n\n**Don't compete to win. Compete for sportivness and learn** \n- Many of kagglers aren't behind medals. That are just rewards kaggle recognises for a user. For many kaggle is a hobby. A late night fun and for others its a project. In the end .00000X accuracies doesn't matter, what matters is what you learnt from this competition. Take away the good. Don't repeat mistakes of past. And when you attend any new competition again, you are a Kaggler! \n\n*Attaching a scene from Avengers. I felt like that at start*\n\n![](https://i.redd.it/ay5cfurj0f301.jpg)", "453174": "Hi everybody, I'm a newbie in Machine Learning and in 'Kaggling'. And I want to ask experienced kegglers and data scientists to give some pieces of advice on how a newbie can improve his skills?\nI know that it is a lot of posts throw the internet about it, but it will be cool if some guys with real experience share the list of technologies they commonly use, tasks that they usually work on. \nIt will be perfect if someone will share links on useful things, because of some kind of specific papers usually hard to understand, official documentation can't give all answers and so on.\nAlso, it will be cool to get some advice about the best ways of 'kaggling')) Everyone wants to take a medal but I read, that kaggle, first of all, it's a platform for learning and sharing (I want to say very big 'Thank You' for all guys who make awesome kernels and answers in discussions!!)((\n\nP.S. sorry for my English((", "453720": "I'd recommend to join our upcoming session of http://mlcourse.ai - open ML course by OpenDataScience. This time, we'll have 4 competitions (Kaggle Inclass and other platforms) during the course. We'll provide benchmarks and step-by-step guides how to beat them. The previous announcement is here on Kaggle https://www.kaggle.com/general/68205 though it's a bit different, the new session is much more competition-focused.\nps. considering that you can read Russian: this article is exactly about kaggle best practices https://habr.com/company/ods/blog/426227/", "455547": "I am also a newbie in this. In my course curriculum, i studied data analytics in my postgraduate program in industrial engineering and management. From there my journey started. i got more interested in this topic, then my professor told me to start learning R or python, It will be helpful for doing analysis. After the end of this course my professor suggested me a course on machine learning by Andrew Ng on Course. I have completed this course and now am learning deep learning by the same professor on coursera. From kaggle also i am learning. I am participating in the competition and improving day by day. You also follow the same path.", "453338": "Hi Vladyslav, since almost everyone has a different definition for newbie. Could you please share your background on machine learning ? (what have you learned up until now, so that some experts can understand you better, and they can give you appropriate directions)", "453197": "I'm also a newbie but this is a one of the great courses of machine learning and deep learning :\nhttps://course.fast.ai/ml.html" } }