Shivam Chhuneja’s Post

OSM is quite a repo to look at the data from! 😉

𝑻𝒓𝒖𝒔𝒕 𝒚𝒐𝒖𝒓 𝒏𝒖𝒎𝒃𝒆𝒓𝒔? 𝑻𝒉𝒂𝒕’𝒔 𝒄𝒖𝒕𝒆. 𝑯𝒆𝒓𝒆’𝒔 𝒘𝒉𝒚 𝒚𝒐𝒖 𝒔𝒉𝒐𝒖𝒍𝒅𝒏’𝒕 👇 I was going through the Mathematics for ML specialization on Coursera(Imperial college one) and found this example in the multivariate calculus course. A classic example in the image, Anscombe’s Quartet. On paper, all four datasets have identical statistical properties: same mean, variance, correlation, and even the same regression line. But when you visualize them? They’re telling completely different stories. 📊 As a data scientist we should know that statistics can lie, and models can mislead if we don’t look deeper. Fancy algorithms and statistical summaries are great, but they aren’t enough to truly understand the data. Here’s why this matters for anyone working with machine learning or regression: 1️⃣ No matter how accurate your model seems, you need to visualize your data. A perfect regression line on paper can be way off in reality. 2️⃣ Your data may look good by the numbers, but without understanding the patterns and nuances, you risk missing critical insights. 3️⃣ Even the best machine learning algorithms can’t fix bad data. Understanding the story your data tells is key to building models that actually work in the real world. (Garbage in, Garbage out) How often do you visualize your data before trusting the results? #DataScience #MachineLearning #Regression #DataVisualization #Statistics #AI #DataDriven #Coursera

𝑫𝒂𝒕𝒂 𝒔𝒄𝒊𝒆𝒏𝒄𝒆 𝒊𝒔 𝒎𝒐𝒓𝒆 𝒂𝒃𝒐𝒖𝒕 𝒂𝒑𝒑𝒍𝒊𝒆𝒅 𝒔𝒕𝒂𝒕𝒊𝒔𝒕𝒊𝒄𝒔 𝒕𝒉𝒂𝒏 𝒂𝒏𝒚𝒕𝒉𝒊𝒏𝒈 𝒆𝒍𝒔𝒆. 📊 Tons of courses and people talk about just programming when talking about data science and leave out stats, quoting "math is taken care of with python" Well, if you’re diving into data science and think it’s all about coding, think again. Mastering statistics is where the real magic happens: because at its core, data science is about making sense of numbers, trends, and patterns. A great advice that Sundas Khalid had shared in one of her YouTube videos as well, here is the link to the vide(it will start at this exact timestamp): https://lnkd.in/dYqD-ug8 Data Science is more applied statistics than coding, and knowing how the math works allows us to think much better about what specific types of analysis would fit the specific dataset we might be working with at that time. If you’re serious about growing in this field, don’t skip out on statistics. That’s where the real value lies. #DataScience #Statistics #Mathematics #DataDriven #MachineLearning

𝑻𝒉𝒆 𝒋𝒐𝒃 𝒎𝒂𝒓𝒌𝒆𝒕 𝒓𝒊𝒈𝒉𝒕 𝒏𝒐𝒘? 𝑻𝒐𝒖𝒈𝒉 & 𝑪𝒓𝒂𝒛𝒚. Especially for anyone looking to break into tech. And it’s not just the fresh grads feeling the heat - competition is coming from every angle: new graduates, experienced professionals, people transitioning from other fields… and everything in between. I’ve been on a unique journey myself. My bachelor’s was in Electronics Engineering, then I did a Master’s in International Business, worked in growth marketing, and now, here I am, learning ML & Data Science. I didn’t realize how much the fundamentals I learned earlier in my career, in areas like engineering and growth hacking, would come back around to help me today. It’s all connected. So thought of sharing some strategies that are working for me and might help you too in this chaotic market: 1️⃣ 𝐋𝐞𝐯𝐞𝐫𝐚𝐠𝐞 𝐖𝐡𝐚𝐭 𝐌𝐚𝐤𝐞𝐬 𝐘𝐨𝐮 𝐔𝐧𝐢𝐪𝐮𝐞 You don’t have to follow a linear path. In fact, the more diverse your skill set, the more you can stand out.(unpopular but that's my opinion) I’ve found that combining my engineering background with my growth marketing experience is a massive advantage in the data science. Look at your own background - whether it’s in a different field or a non-tech degree - and find a way to show how those skills apply to the role you’re aiming for. 2️⃣ 𝐁𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐀 𝐏𝐞𝐫𝐬𝐨𝐧𝐚𝐥 𝐁𝐫𝐚𝐧𝐝 Think of your combination of skills as a product that needs marketing. A great concept that phyl terry talks about is "Candidate Market Fit" With a combination of skills we also need to start posting about our projects, lessons learnt, and even our struggles. Don’t wait until you “know everything” - share your journey as you learn! 3️⃣ 𝐁𝐞 𝐒𝐭𝐫𝐚𝐭𝐞𝐠𝐢𝐜 𝐢𝐧 𝐍𝐞𝐭𝐰𝐨𝐫𝐤𝐢𝐧𝐠 We’ve all heard it, but networking is truly the game-changer. It’s about the relationships we build and the conversations we have outside of job applications. 4️⃣ 𝐏𝐢𝐜𝐤 𝐚𝐧 𝐀𝐫𝐞𝐚 𝐚𝐧𝐝 𝐃𝐢𝐯𝐞 𝐃𝐞𝐞𝐩 As I’m learning more about ML & DS, I’ve found that being a generalist won’t always cut it. Specialize. The market might be tough right now, but the opportunities are there if you play to your strengths and position yourself well. #DataScience #GrowthHacking #CareerJourney #CareerTransition #PersonalBranding #Networking #DataDriven #AI #CareerAdvice

Running analysis on open source repositories and understanding what is going on with them when it comes to DevOps has been an eye opening experiment for me personally. There is so much exciting stuff going on under the hood including but not limited to super active contributors, quite a different balance of type of work getting focused in these repos etc. Love it! #opensource #dataanalysis

I spent the weekend diving back into the math behind machine learning -specifically through Coursera Machine Learning Specialization(Imperial College London one). 📚 ps. I hear DeepLearning.AI Math specialization is really good too. Anyways, what’s funny is that going back to the basics has become not only useful but actually fun. As I progress further into my learnings in data science and machine learning, I’m finding that revisiting these fundamentals is like unlocking new levels of understanding. Things that didn’t quite “click” the first time around are now making perfect sense, especially with some hands-on experience under my belt. Every time I revisit core concepts, I pick up something new. And honestly for me, it’s a reminder that no matter how advanced you get in a field, the basics are always worth revisiting. Link to the specialization: https://lnkd.in/d_Ss6Mbk Would love to hear your thoughts or experiences! #DataScience #MachineLearning #Math #MultivariateCalculus #AI #GrowthMindset

"The story of AI is the story of Compute! People still underestimate the role that compute has played in the development of where AI is today" 🔥 Going through this next level conversation on Andreessen Horowitz YT channel with Fei-Fei Li and Justin Johnson! Highly recommended video for anyone interested in anything to do with data, AI and tech IMHO 😀 Here is the link: https://lnkd.in/duFYVexv #ai #machinelearning #deeplearning #spatialintelligence #tech #data

7 must read books to become a world class data scientist 🫡🔥 These are the books I am going through along with my masters in data science curriculum to build a much deeper understanding of the science behind data science ❤️ #datascientist #datascience #datasciencetraining #datasciencecourse #selfstudy #machinelearning #techjobs

Very interesting episode on Together AI with Chip Huyen! Big fan of Chip's approach to ML design and her blog articles as well. A highly recommended series of hers is her blogs on building a career in ML too! Interview link: https://lnkd.in/dky7PhWN

👩💻 SQL Interview Question(Especially Marketing Analytics!) 🚀 𝑾𝒉𝒂𝒕 𝒊𝒔 𝑼𝑵𝑵𝑬𝑺𝑻 𝒇𝒖𝒏𝒄𝒕𝒊𝒐𝒏? If you’ve worked with data exports from tools like Google Analytics, you’ve probably encountered nested data. (it's an easy one but could surprise you if you haven't worked with it) Here is how I'd frame the question in an interview when hiring a marketing analyst: You have a dataset of events from Google Analytics, and each event includes nested columns with details such as event_url, referral_url, and click_text. How would you “flatten” this data? 💡 𝑬𝒙𝒂𝒎𝒑𝒍𝒆: Let’s say we have a table with a nested event_details column like this: 𝘦𝘷𝘦𝘯𝘵_𝘥𝘦𝘵𝘢𝘪𝘭𝘴[{𝘦𝘷𝘦𝘯𝘵_𝘶𝘳𝘭: ‘𝘶𝘳𝘭1’, 𝘳𝘦𝘧𝘦𝘳𝘳𝘢𝘭_𝘶𝘳𝘭: ‘𝘳𝘦𝘧1’}, {𝘦𝘷𝘦𝘯𝘵_𝘶𝘳𝘭: ‘𝘶𝘳𝘭2’, …}] To flatten this and analyze the event URLs separately, we’d use: 𝘚𝘌𝘓𝘌𝘊𝘛 𝘦𝘷𝘦𝘯𝘵_𝘥𝘦𝘵𝘢𝘪𝘭.𝘦𝘷𝘦𝘯𝘵_𝘶𝘳𝘭, 𝘦𝘷𝘦𝘯𝘵_𝘥𝘦𝘵𝘢𝘪𝘭.𝘳𝘦𝘧𝘦𝘳𝘳𝘢𝘭_𝘶𝘳𝘭 𝘍𝘙𝘖𝘔 𝘦𝘷𝘦𝘯𝘵𝘴 𝘜𝘕𝘕𝘌𝘚𝘛(𝘦𝘷𝘦𝘯𝘵_𝘥𝘦𝘵𝘢𝘪𝘭𝘴) 𝘈𝘚 𝘦𝘷𝘦𝘯𝘵_𝘥𝘦𝘵𝘢𝘪𝘭; 🔍 𝑾𝒉𝒂𝒕’𝒔 𝒉𝒂𝒑𝒑𝒆𝒏𝒊𝒏𝒈 𝒉𝒆𝒓𝒆? The UNNEST function explodes the nested array, giving each item its own row, so you can now work with each event URL and referral URL directly. Have you used UNNEST before? When? #sql #dataanalytics #marketinganalytics #analytics #mysql @sqlquery #sqlinterview #leetcode