Machine learning notes - the story of master kaggle Janio Martinez Bachmann

         When I have to code next to someone , I tend to feel nervous

        Janio Martinez Bachmann My life can be summed up in one sentence ：“ Mom ho ho ho , aww ！” I like playing Mario on Nintendo in my spare time Kaggle The master , stay Voodoo.io As a data analyst .Janio From the Dominican Republic , He holds a graduate degree in financial planning from Humber College, Canada .

         In an interview with Analytics India Magazine In an exclusive interview with , This financial analyst turned into a data analyst and shared what he became Kaggle The story of the master .

         How did your fascination with algorithms begin ？

        Janio Martinez Bachmann：

         Most of my experience comes from the financial industry . I used to work in the credit bureau of the Dominican Republic . My daily tasks are highly dependent on Excel Tools such as . Although I do like my job , But I always ask myself if there is a better way to accomplish these repetitive tasks more effectively . therefore , I began to delve into Python And other programming languages , And bought a Book ——Aurelien Geron The use of Scikit-Learn and Tensorflow Machine learning —— This has changed my view of algorithms and data science .

         This book taught me how different algorithms work , For example, linear regression 、 Decision tree 、 Unsupervised model （ clustering ） wait . When I started reading this book , Data science is not so much hype . I'm not sure what I'm doing . However , I like mechanisms for how different models work and how they can be used to solve business problems —— It fascinates me .

         What was the initial challenge , How did you solve ？

        Janio Martinez Bachmann：

         I must be honest . I was not a master of mathematics in high school or college . One of the most difficult challenges I faced was to understand how the model worked . It feels like entering a dark room without a light bulb .

         However , My curiosity about different algorithms prompted me to understand how these black box algorithms work . therefore , I began to pay attention to many Youtube channel ;Joshua Starmer It's one of my favorites .

         I remember doing one about DNN（ Deep neural network ） Exercises on how to derive specific outputs . I have to carry out forward and backward dissemination on paper by implementing the calculus concepts learned on the Internet .

         If you ask me now how to deal with these challenges , My only answer is ,“ Just curious ！”.

         You don't need a PhD to know all these things . Curiosity is good enough . My advice to beginners is to enjoy the journey , Don't be intimidated by all the terms ; All these concepts can be learned from the Internet .

         What excites you most about coding ？

​        Janio Martinez Bachmann： My favorite thing about coding is that you have unlimited possibilities to complete your work . As a data analyst , I often face the challenge of finding insight , These insights will enable my employer to take advantage of opportunities in the market . however , When dealing with large amounts of data , How do I provide holes ​​ Insight ？ The beauty of coding and open source packages plays a role here . Coding ability is like having superpowers ！ When you know how to code with different tools , The possibilities of how to solve the problem are endless ！ This is what I like most about coding —— It brings creativity and efficiency in solving everyday problems .

​         How did you get into this area ？

        Janio Martinez Bachmann： Believe it or not , I encountered a lot of difficulties in entering the area . Now , Distractions come from all sides , It's hard not to be distracted . however , When you need to pay attention to details （ Coding is common ）, It's important to stay focused . that , What is my daily life like ？

         First , I hid my cell phone away from my desk to get into the area . Why should I do this ？ My cell phone is my main source of distraction , Because I tend to get constant notifications from there , And the closer I get to my cell phone , The more I want to see what that notice is . therefore , To avoid this temptation , I usually put my mobile phone in a hard to reach place on my desk .

         I am an early bird . The first thing I do is prepare my daily to-do list . This gives you a better idea of what should be done during the day , Let you have a better sense of direction . For me, , There is nothing worse than starting a new day without knowing what you will do . I feel completely lost . Once I have my to-do list , I will feel that I have a sense of mission during the day . My daily to-do list will be the first step before entering focus mode .

         What is your machine learning tool ？

        Janio Martinez Bachmann： My most common tools include ：

        SQL（ Structured query language ）： I mainly use it to extract all the necessary data directly from the database . ad locum , I perform the necessary transformations , So as to be analyzed or passed later BI The tool displays this information .
Tableau： talk about BI Tools , This is the dashboard I currently use to present all the necessary insights to stakeholders . There are other platforms , Such as PowerBI、Looker、QlikView etc. .

        R & Rstudio： I mainly use R To perform statistical analysis and A/B Testing process , But there are other functions , For example, data conversion 、 Visualization and so on .

        Python： I tend to use Python Automatization is often a repetitive process .

        Shiny Web Apps： I use them as a kind of dashboard . The only difference is that it is more flexible to integrate machine learning models into Web In the application .

        DBT（ Data building tools ）： This is the latest tool I am currently learning , But it will change the rules of the game , I would say that it must be learned in the foreseeable future . It uses software engineering principles to transform 、 A tool for testing and recording all forms . I currently associate this tool with Redshift Use it together .

        Git： This is a tool that anyone needs to learn in the end , Because in most organizations , You need to work with your code . By understanding Git command , You will be able to use Github、GitLab、Bitbucket And more collaboration tools .

         How to prepare for the first hacking Marathon ？

        Janio Martinez Bachmann：

         In my hack Marathon , I mainly use Python To solve the problem . therefore , My suggestion is to start there , Because it is the most commonly used language I saw in the Hackathon . however , In machine learning library , I suggest learning Pandas、Matplotlib and Scikit-Learn The basic knowledge of and the concept of circulation , So as to have greater flexibility in manipulating data .

         What is your biggest dissatisfaction with the hacker marathon ？

        Janio Martinez Bachmann： honestly , When I first took part in the hacker Marathon , One of the main challenges I face is working with others . It's not that I don't like working with others . When I have to code next to someone , I tend to feel nervous . When you have to display code or work on the screen , Have you ever been in a coma ？ ok , This happened to me in my first hacking Marathon .

I worry about what other more experienced programmers will think of my skills .

         however , We should remember , We are not born coders . therefore , My advice is , Don't be afraid to take part in the hacker marathon . See this as an opportunity to learn from people with more experience in the field .

         As a programmer , What was your worst experience ？

        Janio Martinez Bachmann： As an analyst , I often interact with other stakeholders , To visualize the needs of these stakeholders . As an analyst , One of the worst experiences is that you have to deal with a stakeholder who asks you but doesn't know what they want . In the working environment , This can be frustrating , Because you think you have to guess in some way what that person wants . Fortunately, , There are some techniques that can handle these situations , One way I suggest to implement this is to keep asking questions . By asking questions , You will be able to define the problem , This will allow you to specify how to resolve a particular problem or request .

         Another bad experience I have had is , When I design a project for a large number of stakeholders , Only a few people use them . This has been frustrating and frustrating , Because some stakeholders may ask for something under a sense of urgency , Make you feel that stakeholders need this . However , Only a few people find the final project useful after it is completed . This has happened to me several times , Especially when making dashboard . To solve this problem , I'll go back and ask questions ！ most important of all , Ask if the project is necessary and how it will affect the organization .

         What attracted you to join Kaggle？ up to now , How was your journey ？

        Janio Martinez Bachmann： When I started reading Aurelien Geron Of “ Use Scikit-Learn and TensorFlow Practice machine learning ” when , I heard that Kaggle. The previous pages mentioned Kaggle . I am curious about what this website is , When I first saw it , I was fascinated by it ！ Why? ？ As a beginner of coding , This platform is very suitable for applying the theories I have learned from reading books . There is nothing better than learning to encode while exploring some data sets and getting stories from specific tables .

         The data storytelling part is driving me to Kaggle One of the reasons , most important of all , There is a great community here that can help you . Learning from a talented person's notebook enabled me to improve my coding skills and learn different machine learning concepts . As for my journey , I have to say it's hard , But it's worth it . I left recently Kaggle, Mainly because of my present job . But I plan to Kaggle Contribute to helping the community .

         Your first time Kaggle What is the game like ？

        Janio Martinez Bachmann： as far as I am concerned , The first competition I participated in was to predict house prices . It was an interesting game , Because this is the first time I have heard of feature engineering （ It is mainly a concept that we extract insightful features to enhance the predictive power of our prediction model ）. Besides , This competition let me learn the interesting concept of advanced linear regression that I had never heard of before . For all that , You can guess I didn't do well in this game , Because this is my first game . But I learned a lot , That's the most important thing ！ So don't be afraid to compete ; They can be fun ！

         Become Kaggle What does the master feel ？

        Janio Martinez Bachmann： I was shocked . Remember 2021 year 3 I was on holiday in the Dominican Republic in August , Lying on the beach , received Kaggle The notice of , I became Kaggle The master . I can't believe , But at the same time, I feel happy about it ！ After four years of hard work , I became Kaggle The master . That doesn't mean you have to wait four years to become one of them . I have seen other Kagglers He became a master in two years . For all that , When I received from Kaggle When the news of the , I am still ecstatic ！

         stay Kaggle Winning skills in the competition .

         Create content ： When I say create content , I mean explore datasets that only a few people have explored and that you might find attractive to the community . When I discussed an interesting topic when dealing with unbalanced classification , I can give you an example . At that time , The topic is Kaggle Not in China “ happen ”, So I decided to take this opportunity to create a center around “ Credit fraud || Dealing with unbalanced data sets ”. It took me three months to create this notebook , But it's worth it , at present , It's close 4k Like .

         Participate in discussion ： If you want to Kaggle The community promotes your brand , I suggest you join the discussion section , There are two main reasons . You will learn about other topics through many discussion topics Kaggler, And learn with them in all these discussions . This is a good way to make yourself known in the community .

         Respect the community ： When I say this , Please try to behave ethically in the whole community . I've seen some unethical behavior promoting your notebook on different notebooks , So that people like your . however , I advise you not to do so , Even if you might be tempted to do so . One , When someone directly asks to do so , Other users won't like it , Two , This may seem a little unprofessional , This will damage your reputation . That's why creating content is important , But most of all , Enjoy the journey ！ Whether you are a master or a master ; It is important to , You are in Kaggle And other fascinating communities ！ Want to have patience ！