How I Learned Data Science | My Data Science Journey

I'm going to discuss how I learned well learning data science because I think learning is an ongoing journey it will be wrong to say that I have learned it and I'm going to discuss my whole journey and in the end I'm going to discuss the two most important skills in data science journey so please make sure you watch till the end let's begin with my background. I have a bachelor's in computer science and I have been working as a software engineer for last 15 years so my main role has been that of a software engineer so I don't have a formal education in data science I don't have any statistics or math degree but i'll tell you how i picked up my interest so seven years back when I was working in my software engineering team we were using mainly pearl and shell script for automating you know small tasks within our team at that time python started becoming popular so I tried python for one or two scripts and I fell in love with it because of the very easy syntax that it offered so compared to paul or shell script it took very less time to automate a task using python so I started automating small tasks at my job using python so python is great with with this kind of stuff if you're in college or if you're working in any role there is a scope of using python to automatically automate your day-to-day tasks okay so I started learning python and after that I picked up something very important uh very important tool that any data scientist will use and do you have any hint on what that tool is by the way that thing is in the background somewhere yes you guessed it right that's pandas so pandas allows you to do data cleaning data exploration and I learned about panda from some blog on internet I started learning pandas myself started building small notebooks around it and then I introduced pandas in my team so I told my colleagues that okay look this is a great thing this is a great tool so why don't we start using it and then we started using it at my job and in parallel I started making videos on pandas they say that the best way to learn something is to start teaching it and i'm a huge believer in that so I started teaching pandas and that consolidated my own understanding because what happens is when you post a video someone will ask a question on pandas and then if I don't know the answer I will figure it out i will respond to that answer so it kind of consolidates your understanding and it makes the whole concept more clear basically so this way of teaching promotes a critical thinking almost right and because of that my own understanding about pandas became more and more strong as I practiced on more like kegel data set I made more videos and things like that and there is a quote by Einstein saying that if you cannot explain it simply you have not really understood it so I was always like okay I want to explain this concept in a most precise clear way and that will confirm my understanding about my own topic so I consider myself a lifelong student you know on youtube channel when I'm posting videos people call me sir but I'm really a student I spend one to two hours daily almost daily to learn something new because there are so many exciting things happening in the field of data science so I just constantly learn new things and as i said that teaching promotes critical thinking let's say I'm explaining gradient descent to a new student now we talk about loss mean square error and students usually ask me why do we use mean square error why why can't we take a simple absolute right and at some point I did not know that so I googled it I talked with my friends and I found out that by using mean square error the gradient will converge smoothly you know so and then you learn about convex function okay now what is convex function I don't have a degree in math so i will ask this question to my best friend and you know who is my best friend everyone please repeat with me Google is my friend google is my friend google is my best friend so if you have a question ask google so then what is convex function and google will probably tell you give you the answer so so I learned math and test statistics that way because if I had spent time just learning math and statistics for less six months it would have taken so much time right so all right so I learned python then I learned pandas then I started andrew Ang's famous machine learning course now probably everyone is aware about his course he's very popular in the field of machine learning his videos are available on youtube. I think and on coursera also that course is available for free so I went through that course and as i was going video by video let's say I learn about linear regression now in linear regression you know what is linear regression really and what is greater and decent and how what is cost function what is mean square error so all these concepts I will learn by referring to various resources so for math a can academy is an excellent course there is a Youtube channel called three blue one brown math is fun.com is another excellent resource mathisfun.com is like explaining math to kids you know very simple exam examples they use and it's so good I love that website so i would go through andrew hanks let's say decision tree video and if that machine learning concepts require me to get clarity on five different math and statistics concept then i'll pause that video I will go to internet you know go through those concepts and then learn those concepts and then come back so it's always like here is your main fork your learning machine learning then math statistics you have to always come out learn math and statistics go back then again machine learning come out again learn math and statistics so I continued that journey now someone asked me this question how do I keep myself motivated well see the idea is if we are producing something tangible after learning a concept that will keep the motivation high for me it was youtube so if I learn some concept I make a video on it okay video gets a lot of views and those going up like those views going up it's kind of a mental reward you know it generates this brain stimuli almost and people ask question I respond to that question so I feel like I learned decision tree but I'm making some good use of that it's not like I learned mission decision tree and that's it nothing happened you can build small solutions for local businesses or build some small products where you can use your knowledge and when you see your knowledge being transferred into a tangible product that is useful to someone that will keep the motivation high okay and I always take one thing at a time in a small small step and don't worry about too many things because it might be overwhelming machine learning and data science is so vast sometimes people like are like oh and am I going to learn it well take a small step every day okay and over a period of time you build your knowledge then chem statistics statistics is equally important and for statistics I have another great youtube channel that I refer to and you probably know about it bam steadquest statequest is the channel I refer to for statistics and there is a book called practical statistics for data scientists that I'm reading you know I keep on reading keep on referring to that book so there are some books some resources available for statistics that I always you know on ongoing basis I refer and get my understanding clear when I'm learning pandas or machine learning I always believe in project based learning so I refer to kegel for different data set with different challenges you know they they post different data sets and you can also look at notebooks created by other people so you kind of get different ideas on data exploration how to build machine learning models and so on so i kept on practicing that on my free time and then I started building small projects so data science projects if you look at those projects on my channel I have built two different projects one is let's say Bangalore property price prediction website so I build an end-to-end website with machine learning deployed to amazon cloud and so on and I learn things by practicing those things on my own and then i share with people on youtube so again going back to learning by teaching concept uh i started building those projects the other project was a spot celebrity image classification so that I learned about you know computer vision and things like that so that way by building those projects uh i shared my knowledge with people and at the same time I consolidated my own understanding now I mentioned this before that every day one to two hour I spent learning something new and I made a different video on how do I stay updated in the field of data science but I'll quickly summarize it so I attend two or three conferences every year I don't spend money in upgrading my phone or buying some expensive items but I make sure I spend money in attending conferences because for me conferences and network building is a very very crucial thing for my career so I attend jupiter con pi data ai events such as gtc. gtc was free this year nvidia it is a con biggest ai conference by nvidia and in those conferences I will also attend workshops for example in pi data I attended nlp workshop in gtc I attended deep learning workshop so by going through this workshops you get to learn something new you know i have that curiosity that I want to learn something new so we need to have that student mindset like never think that I'm done studying something and because of that I get to know about new concepts every day for example I learned about rapids and ask data frame from a pi data conference okay so i would learn a new term and then I will come back home I would Google it I would try it out on my own and that way i build my knowledge on a given topic and the best thing about conferences is the networking you get to meet data scientists and machine learning engineers working for top companies and I make sure I add them in LinkedIn during that conference so I build that network which is super useful for my my career growth and then I read some blogs as well such as towards data science analytics video dot com I attend ml seminars and my job again that allows me to learn new things I follow certain twitter handles such as the twitter handle by person called elvis and I'm going to provide all these links in the video description below by the way then I follow LinkedIn post you know I'm connected with data scientists machine learning engineers around the world so if you also connect yourself then if you spend some time every day on linkedin you will also learn a lot of new things and then there are live interactions that i'm doing on my youtube channel with data scientists throughout the world and when i'm doing that interaction I almost always learn on new topic or a new thing thing from other people podcast is another great way data frame super data science are the two forecasts that I follow again they invite the industry leaders in the data science and AI and you know by listening to those discussions let's say you're driving to work you can listen and listen to those podcasts in the audio format rather than wasting your time on you know just surfing internet facebook or watching news you know it's not very useful i mean many days i watch news you know after like two three days i i sometimes have news fast day where i don't consume news at all and that has helped me a lot personally now let's discuss the two most important skills in my opinion that any data scientist need to have and how i am learning those skills number one skill is business understanding and analytical thinking as a data scientist you know you're solving a business problem at the end of the day and if you don't have a good business understanding there is no point of using your skills right so I'm in a finance industry and I always ask meaningful questions for finance business to my colleagues I have my colleagues who are expert in finance field and i'm always curious about you know various things which are going on in finance industry and these are not technical things these are like you know you call it a domain knowledge i also do google and try to build my understanding in finance because i have been in this domain for last 10 years now you could be associated with any domain the good thing is data science is applicable in any domain right so if you are in a petroleum engineer or a mechanical engineer there is an applicability of data science in that domain and having a solid domain or business understanding is a very important skill that a data scientist need to have so if you are in your field you are talking with your colleagues you are talking with your business partners always have this curiosity and always try to think about what pain a business or business owner is going through and how these tools you know python machine learning these are just the tools and these tools solve a given problem so given problem is very important the understanding of that problem is important and then you can apply you can open the toolbox and use one or the other tool to solve a particular problem so building a domain knowledge is very crucial and most of the time I have seen people don't focus on that much people focus like 90 to 95 of their time in technical skills and that's the biggest mistake now sometimes people say okay how do I build domain knowledge well you build it by doing it every day you need to have that curiosity you need to be asking that question to right people and that will build the understanding over a period of time it's not gonna happen magically so that's number one skill number two most important skill for a data scientist is presentation and storytelling okay I was reading this howard business review article where someone asked a question to a prominent data scientist in usa and the question was which skill is more important for a data scientist is it building a deep learning model or making amazing powerpoint slide and that person answered saying that powerpoint slides yes that making a good effective presentation and storytelling is such a crucial skill and unfortunately not many people spend time on this now how do I build my presentation skills well number one I'm always looking for an opportunity to present or to give a seminar so in my team when I'm talking with my colleagues you know I'm always eager to present so if I see a little opportunity I will enroll myself you know I would volunteer even I go to my manager and say okay can I speak on this event can I give a talk during this particular event and because of that when I create those opportunities for myself that gives me a platform to improve my presentation skills and my communication skills in general you know I try to communicate my ideas you know very few words in a most precise way and many times I would ask my colleagues who are listening to that presentation to give me a critical feedback at the end of the presentation so let's see if I'm presenting on some idea I would tell my my good friend that okay give me a totally blunt honest feedback on how I did during that presentation and that has helped me a lot because I get a feedback such as you use a lot of filler words or did I make a transition smoothly from a problem statement to a solution because problem statement is very important it's like if i'm talking about tensorflow serving i need to first explain what is the pain what is the pain that tensorflow serving sauce it's like making kid hungry before feeding him or her the food right so spending initial few minutes few seconds on a problem statement a pain point is very very important then we transition smoothly from pain to a solution in a very effective manner I like to use visuals in my presentation there is a great talk on Youtube called death by powerpoint I strongly suggest you leave all your python deep learning courses everything besides see those things are secondary the most important thing is presentation business understanding so you watch this video death by presentation you will learn amazing skills that can be helpful my friend moiselli recently retweeted in lingering saying that you know I'm going to show that post here see CEOs make most money and what is the tool they work on ceos don't code in python they work in powerpoint so now you understand the importance of powerpoint and presentation in general these are the soft skills that people don't pay too much attention to well I had if you found this useful please share it with your friends and if you have any ideas or any interesting stories on your journey please post in a video comment below I'll try to respond to that comment and thank you very much for watching.

Channel: codebasics

Views: 67,274

Keywords: yt:cc=on, how did i learn machine learning, data science learning, data science learning path for beginners, data science learning plan, data science, data science for beginners, data science for beginning, data science journey, codebasics data science, learn data science, how to learn data science, how i learn data science, data science codebasics, code basics data scientist, how i learned data science

Id: -LVdJuNzmU0

Length: 19min 31sec (1171 seconds)

Published: Mon May 24 2021