All You Want To Know About Kaggle- Podcast With Abhishek Thakur Kaggle Grandmaster- Give Away Books

[Music] okay hello guys so we have Abby shake with us and if you don't know about abhishek guys is also a data scientist he's is having an expert in NLP that is natural language processing and he's also a CAG ler he is basically the cattle grandmaster so we'll just try to get more information about him and we'll have a quite a discussion with respect to cattle in this particular session so welcome abhi shake probably most of our viewers know you the pretty much famous your most favorite among you know cackling people know you with respect to that everything that you want to introduce apart from this you can actually introduce yourself with respect to that sure attrition first of all thank you very much for inviting me for having me here and I must say that I'm also a plan of yours so thank you very much for all the cool videos that you're making so I follow them from time to time and I'm yeah I'm a data scientist so I'm currently based in Norway I I was born and brought up in India and since last ten years I've been living in Europe so the plan is to come back to India at some point yeah and when I'm not working so that's the time when I'm when I'm doing all different kinds of stuff like writing a book and being on Kaggle competing with other people on cattle so in different competitions and that's where most of the things that I've learned flown from so okay you won't believe you won't believe I have a lot of telegram channel I just announced yesterday that I'm going to have some session with avi shake then they were like so just make it quickly you know just upload the video you want to see we want to see we want to see so you have a great fan following here in India also now considering that we shake how important is skagle for becoming a data scientist will it help in their journey or in the kind of work that they do in the company well that that's a very interesting and very difficult questions you know and it can get controversial so it totally depends so if you if you're asking me then I would say yeah it's self but if you're asking someone who's already a data scientist and has learned quite a lot in the industry then they might not find it useful why it's useful for me is because whatever I have learned I've learned basically from competing in different competitions so if you're a fresher and you don't have any industry exposure and you want to learn you want to build a portfolio because these times are very difficult you know everyone wants to be a data scientist but nobody knows where to start from and you can do all kinds of certifications all kinds of courses but if you don't have this kind of practical touch then you're lacking you lacking something and I think in the interviews that matters a lot so if you like what I have what I've seen in India it's like you have to be from the top colleges right and if you're not from there and you have to prove that you are still as good as them you have to have a good portfolio and I think kaggle competitions or competitions in general like machine learning competitions or different platform can help you build that portfolio so you have actually told my story itself I'm also not from a you know tire one College itself and from a tire three college I came from as a software engineer background I was a.net develop actually and CagA has played a wonderful role in my life you know just to understand the basics understand the good practices that we can use now I have learnt a lot from catalyst as it is okay now just tell me because yesterday when I posted in LinkedIn you know one of the guys said what is Kaggle he did not know I think you also you also commented below that particular post I guess so just define what exactly skagle for a new user who's coming for the first time in this platform as a data science you know as a data science member to contribute into these communities you just define what is Kaggle in short so basically my answer was its underground community so yeah it is a community of data scientists and there's not just competitions obviously competitions are the main part of Kaggle so you get to compete in different competitions against a lot of different people from different backgrounds so they're physicists they're people who have studied chemistry they're from different backgrounds and they're trying to learn data scientist data science and some of them are really good data scientist already and have worked in industry for many years but many of them are fresher they're new people so Kaggle is this community of data and I think they have over a million members now where you can compete against each other and you can share you can share your views on different things that are going on in competitions or you can share code so you can help others learn and they also have a learning platform where you can learn data scientist data science and applied manner ok so now I'd really like to know what was your path you know when you entered into kaggle to becoming a Grand Master so what are difficulties you faced how did you go ahead with it and how did you achieve it so whenever I'm talking to a lot of people were new in gaggle and they want some kind of tips and tricks so I was like them so I also needed some tips and tricks and when when I had joined gaggle it was I think back in 2013 although my profile says I joined nine years ago but I started in 2013 and at that time I was I was doing something entirely different from machine learning and I wanted to learn machine learning and I saw this platform ok nice there is an image completion going on I like images let's start with this so start building something basic and I I was obviously from the bottom in the leaderboard and then I saw the solutions from others and I started looking into new terms ok what is random first I don't know ok random first started with random forest I came across decision trees so going basic and looking into different terms on Google or YouTube and looking for tutorials and videos and also reading some research papers so that's how I began and I think now nowadays if you if you have you have so many resources to learn this theoretical kind of things right all kinds of theory on different channels and then once you go around applying you once you have you have to find a project and you have to apply what you have learnt so I think that's the best best place to do that is doing some kind of machine learning completely so I'm not saying go to kaggle and go to Titanic competition because that's already been done several times so try something new and obviously you will fail a lot of times but you can learn from people who succeed so definitely it's like I said I have referred many people Colonel you know and based on that I have learnt a lot of things with respect to the lifecycle of a data science project I feature engineering feature selection and many more things are such so this is awesome and just tell me like there's one misconception that whether a single person will be able to crack a cattle competition or they need to be a group of people you know because I have seen many group of people will also be participating in Kaggle itself right three to four people so what do you think about that definitely you can you can a lot of people have competed individually and have succeeded a lot of times so there's nothing new about that but when you're in a group probably you have some advantage sometimes you also have this advantage like you you see a group of five people but probably only one of them is working in that group so that's the disadvantage of group but when you have a good team then yeah it's much better to learn from others if you don't have a team if you want to be a grandmaster you have to get so there are different levels on guys all right and if you have to reach the Grandmaster level you have to get a gold medal without any team so that's also very important if you want to reach there you have to do this thing and I have seen a lot of competitions even recently in which individuals have won so that's generally something people have been doing for a long time and they will continue doing but see if you are ranked 20 or 25 and you see like most of the teams in front of you our team's actual teams or groups and you are doing it individually and still you rank 2095 I think this it's a very good rank and it's like if you write about this in your resume I think it will matter a lot so now you share the partner you do the sharing part of it now one thing that I really want to ask now they are various medals also write gold medals other type of metal silver medal now how do you get this kind of medals so at a specific rank you have to come then only you'll be getting it with respect to various competitions or how it is yeah so not sure of the actual formula but I think it depends on the number of people competing in it in competition so you have a certain category of people who are on the top they get gold medals and after that some people get silver and some get prawns some don't get any kind of metal so but I'm not sure of how they calculated exactly and Gordon gold medal that's the like most important like people like it and you know you have achieved it who cares it's it's it's nice - that feeling is nice when you win a gold medal so okay so what I've heard about your machine learning book right that has recently come I definitely need a copy of that you know I need to read I need to see that what you have done and are so yes tell me about your new book what what it actually solves what is what is the kind of column statement how it is different from other books that have already market yeah so first of all I'm I'm very thankful to you we're talking about my book and definitely you don't even have to ask for a copy and so the book is called approaching almost any machine learning problem and this is something that I have learned in the past five or six years so since I started I so I started accumulating a lot of different scripts and putting them on my github in private repos if they're not good and I keep reusing them so I still use something that I've written three years ago and it said this book combines everything that I have accumulated in during these several years and it's a compilation and it's all about applied machine learning so if you know a little bit of fury and you don't know how to start with projects because I've seen that people know about categorical variables Patil people know what categorical variables are but when they have a real problem and they have like a missing value in categorical so they fail so how to have to deal with these kind of problems so yeah everything is totally applied so it expects you to know some theory or just get the book and writing a book called approaching almost any machine learning problem and it's about applied machine learning so when you know a little bit areso i've seen people who would know about categorical variables and but when they are presented a problem with categorical variables most people fail and that's because they have never had any kind of exposure to any kind of applications of this so that's what the book provides and I'm I think I I don't even have any count of how many different kinds of datasets I'm using in the book so for every different kind of thing I'm using a new data set I have also created some fake dataset to go with the problem so yeah I have everything ranging from feature engineering to feature selection and going into images and X data but from applied point of view so that's that's what makes the book different from what we already have and also the price point is different because I'm selling it for $15 which is quite cheap and probably I will reduce it further depending on now if people have people are interested I will just reduce it further so that so that anybody can have it and one more thing I would like to mention that if you if you think like you have some financial problems and you're not to afford $15 then I'm creating a form fill that fill out that form and I'll send you a free copy that's a my I just want everyone to have it yeah that's an amazing initiative it's and I just really want to ask you one thing is that I also have written a book quite long back one year back okay how much time it took actually to write the book so that this will give an idea say which one so I mean most most of it came from like last six years of experience so I I write quite fast if like if everything is going fine in my life and I might write more than 20 25 pages in a day so the book is not huge it's less than 300 pages somewhere between 270 to 300 pages so it did take it did take a while I think I had already announced the book sometime maybe two years ago people don't even remember so that's when I wrote the first chapter of the book and now I'm finishing it so yeah it took a long time if you really good author it cell phone takes a lot of time to write something you know sorry it should because right you should be true true and you should be able to explain it in a very simple way and that also takes a lot of time cold takes a lot of time in these kind of books because you're not writing a story you have a lot of code involved and that takes a lot of time so this is obviously right if you have your own editor if you have your own you know who is actually creating that whole book itself but usually we have different kind of pressure when you are writing for someone else you know so considering this this was amazing itself owner I'm best of luck for your book it said I think you should do something like a giveaway you know so that definitely helped pull through my channel to show any ways that would be possible then definitely so yeah I was I was thinking about that and I think I can probably give away five books to your subscribers and you get to okay perfect so oh do you have any criteria because you may be doing this already it's not here to sell it I mean spiders I'm sorry do you have any criteria to select the subscribers like how should we select the subscribers they just randomly picking up or yeah that's interesting select people who you like most no I mean who are really loyal subscriber of yours subscribers of yours so you can you can do that and also make them subscribe my channel definitely what we will do is that I'll sit along with you who will write the best comment and for this particular video you know will sounds good give them like whoever is the best comment you'll be getting you know will select five people's me and a we shake probably Abhishek will also be seeing this particular video once this particular videos life and you know you can select five people than you can make us so amazing work we shake definitely I would like to see this particular book when you're saying with respect to applied you know AI machine learning you know for a real world problem statement which is pretty much amazing because people face that problem I have seen that also so amazing work how do you feel you know after so many years you are writing the book first of all you know there is something like yes I have to one day I had to do this you know and I mean it's definitely a good feeling but I'm also very scared because I don't know how the audience is going to take it so yeah that's why if you see the description behind me it's very straightforward so I just say like this if this book is for you or this is not for you so be careful about this thing when you buy this book because you're not going to see a lot of basics so I'm not explaining what a decision tree is how large degeneration works but I'm I'm explaining when should you use what so yeah it's it's definitely a good feeling and quite scary yeah this is the lagging you know when they are applying in one scenario that is they're lagging because now everybody probably knows machine learning algorithms in depth and all thanks to our YouTube channel you know what we are doing from so many years definitely I have seen you have seen a lot of videos of yours and they're really nice sometimes I'm like planning to make a video and then I go to your channel and see that you have already done that it becomes very difficult but yeah I really enjoy that nothing just for the data science community not to compete with anyone that is if you also don't know Abhishek also has a YouTube channel again the link will be given in the description just go and have a look subscribe this channel we creators need support from you all because again we are just doing it to contribute to the data science community right Abhishek I guess the truth yeah we already working somewhere you know apart from that we are just taking out this time just to be the part of community and contribute to the community itself so this is amazing talking to you wish I get probably this is the first time we are talking let's see we can collaborate further anything as such you know I come up with life sessions talk to the people that would be wonderful you know when we are actually discussing about various things so any final thing you want to yeah yeah I have a question for you yeah I mean you have given this advice a lot of times but I I just want to ask you since I'm here so what kind of like advice related to cattle or machine learning competitions and other different things would you give to both freshers and people who were working like people who have jobs right now and they want you like switch to data science but related to these kind of machine learning competitions what kind of advice would you give them so I I have actually told this many times in my videos also what I believe the way I have learnt is that to to understand some of the things I've definitely followed cable you know when I am comes to doing the life cycle of a data science project you know whenever I need to fix some problem I need to do something like specifically if I talk about pitch engineering as you said right if category rebels is missing where do I go and see you know that is not be able to get it everywhere taggart problem statements the dataset has this kind of problem so definitely I have referred Carol with respect to that what I'd suggest for the freshers is that initially when they are in the college that is the right time they start you know competing in goggle make a group start competing try to solve different different problem statements that will be helpful in their resume and they'll be able to get jobs internships and all for the people who are experiencing who wants to make transition all the best practices you'll be able to find in Tangled so that once you get those practices you have some experience with respect to working in some IT background some software engineering background it's all after you create a product then you do the deployment you create a model you drew the deployment that is a further process that leads to after you follow everything you can so it is always good to have the combination of both goggle work experience competing work experience along with your industry work experience so I tell everybody don't stock goggle itself learn some other things see how you can deploy models you can deploy models in different different cloud servers that would be definitely yes this was so gaggle and these competitions can only take you to a certain stage and after that you have to do it on your own because maybe on yeah in real-world IT industry what we work right and what we do in Carroll it is completely different true-true it is different but you can you can apply the learnings that you have acquired from these different casual competition to your industrial problem so for me personally it has given me ability to think differently so yeah these kind of things are nice amazing so any last thing before wrapping up the colony motivational thing I usually tell everyone to give some motivational things to the people you know which they require at this specific stage anything that you want to tell um so one of my favorite things is I I have talked about it like almost all the podcasts have been in it's perseverance and hard work so the thing is you should never give up so you might like people going people winning competitions when we talk about computing people winning competitions and people beating you several times going ahead of you but learn from that and don't give up work hard and if you see like if you're scared of someone who is in top ten and you're not if you work hard you will be that someone who is carrying other people's in a few months so that's something you should keep in mind okay so thank you appreciate for giving your time for this podcast and best of luck for your good for a book I hope you do well definitely from this channel I will definitely try to give all the link and information so guys if you like abhishek book please just go and have a look try to support him he has written it in wonderful way I can just see by see seeing the title I can understand a lot of things so thank you Sheikh for giving your time it was wonderful I think the wheels have learned a lot and yes let's see we'll do further collaboration yeah definitely thank you very much for having me here and I'm looking forward to future collaborations thank you so yes guys this was all about this particular video I hope you like it all the information regarding Abhishek YouTube channel book link will be given the description of this particular video yes we'll see you all in the next one you have a great day thank you bye

Channel: Krish Naik

Views: 21,311

Rating: 4.9687228 out of 5

Keywords: data science tutorial javatpoint, data science tutorial python, data science tutorial online free, python data science tutorial pdf, python data science tutorial point pdf, what is data science, data science tutorial tutorials point, data science course, nlp tutorial python, natural language processing python, natural language processing examples

Id: 8wcEHNl8x8Q

Length: 23min 40sec (1420 seconds)

Published: Sat Jun 06 2020