Hustle Hub #28

New to Hustle Hub? Make sure to subscribe for more!

Hey friends,

This week has been super exciting for me because I’m going to launch an online course (Introduction to Python) in partnership with one of the best online coding bootcamps in the world.

This is the first time I’m teaching in the public (live) and I can’t wait to share more with you soon. The best part? You can join us literally anywhere in the world because it’s conducted over Zoom. Stay tuned! 🔥

In today's issue, I’d like to share with you how LightGBM algorithm works, the failure of my my first startup idea and the lesson learned, Python Data Science Handbook, and HuggingChat.

Let’s get to it! 🚀 

🛖 What's in the hub today?

• Tip: How LightGBM algorithm works

• Mistake: My first startup idea failed — prematurely

• Learning: Solve a problem that you care and are excited about

• Book: Python Data Science Handbook

• Tool: HuggingChat

⭐️ 1 Tip

How LightGBM Algorithm Works

As the name suggests, LightGBM is “Light” because it takes less computational memory to run and can give results faster compared to other ML models. It’s also one of the widely used ML models in hackathons, making LightGBM very popular in today’s industry.

So, What’s LightGBM?

• LightGBM is a tree-based algorithm that uses gradient boosting ensemble method to learn patterns from data and make predictions.

• It’s basically like having many different people each make a guess about what the answer might be, and then combining all of those guesses to get a better answer — hence the ensemble method.

• Because of this tree-based algorithm, LightGBM can be used for both classification and regression use cases.

How Does LightGBM Algorithm Work?

• Unlike traditional decision tree models that use Level-wise tree growth, LightGBM uses Leaf-wise tree growth to train its model.

• In simple terms, LightGBM works by repeatedly finding the most important features in the data, and then splitting the data based on those features to create a decision tree that grows leaf-wise.

• This means that given a condition, only a single leaf is split for the next condition.

• But here’s the question, “How does LightBGM know what conditions to split?”

  • It uses a technique called Gradient-based One Side Sampling (GOSS).

  • This technique allows the algorithm to quickly identify the most important data points and ignore the less important ones, which makes it much faster than other decision tree algorithms.

What Are LightGBM’s Limitations?

• LightGBM needs a lot of data to learn from.

  • If there isn't enough data, the model might not be accurate as it can easily overfit.

• It may not work well if the data has a lot of missing values or outliers.

⚠️ 1 Mistake

After I quit my job and embarked on my startup journey, I joined an incubator (Entrepreneur First) and found my cofounder. Our first startup idea was to help F&B restaurants hire part-time workers.

We quickly realised it was really a painful problem for most F&B restaurants (after talking with them) as they constantly struggled with the lack of manpower. We partnered with a Japanese restaurant as our pilot test and helped them find part-timers.

We found the problem, so what’s the solution? 🤔

Our MVP was simple. It was just a website and a telegram channel to invite people to join and incentivised them with flexible working hours and payments.

Even though we managed to help the restaurant find part-timers, we decided to shut this startup idea down. 🙅🏻‍♂️

The reason? It was very operational heavy on our side to find part-timers. We learned that many people didn’t want to join F&B not because of the money, but because of the work nature itself.

And that, my friend, is how our first startup idea failed — prematurely.

🧠 1 Learning

During the same period, there was another startup solving the same problem as we did.

This startup is called Staffie. The founder bootstrapped the startup and made $500k profitably in its first year.

Why did Staffie succeed while our startup failed? 🤯

• Staffie’s founder struggled to find part-time jobs when he was studying in Melbourne. He faced the problem himself, so he wanted to help others find part-time jobs easily while making a side income.

• He fell in love with the problem, not the solution. The solution might change, but the problem always remains the same.

🧠 Here’s what I’ve learned:

• Solve a problem that you care and are excited about — especially when you’re building a startup.

• Building a startup is a very long and tough journey. If you’re not excited about solving the problem or you’re solving a problem that you never had before, you’ll likely give up some time down the road (just like us) due to the lack of internal motivation.

• Founder-Market fit increases the chances of success when building a startup, because:

  • You faced the problem before, so you are more determined and know how to solve your own problem.

  • You have deep experience in the space you’re serving.

  • You have a wide network for you to leverage and land your first few customers.

📚️ 1 Book

Python Data Science Handbook (FREE)

This book is a gem for any data scientist. I read it a few years ago, and it is still very valuable and relevant in today’s use cases. It was practical and packed with many useful codes for you to know how to:

• Clean and analyse data using Numpy and Pandas

• Visualise data using Matplotlib

• Build and fine-tune machine learning models using Scikit-Learn

By the way, this book is free and you can also get all the code in Jupyter Notebooks HERE on GitHub!

📚️ Here are my takeaways from the book:

• How to perform EDA to uncover patterns and insights from data

• How to use various techniques of features engineering to improve ML models’ performance

• How to use cross-validation techniques to evaluate ML models’ performance

🧰 1 Tool

Hugging Face finally made the move to launch HuggingChat as the alternative to ChatGPT.

HuggingChat is built based on the Large Language Model Meta AI (LLaMA), a foundational model with 65 billion parameters from Meta, released in late February 2023.

HuggingChat is also really important because it helps make AI more fair and equal for everyone instead of being monopolised by big tech companies (i.e. Microsoft, Google).

While this is a significant milestone for the open-source AI community, it’s important to note that HuggingChat is less reliable than ChatGPT — at least for now.

For now, you can try HuggingChat on their landing page without having to sign up at all.

🔥 Hustler Spotlight 🔥

👋🏻 How would you introduce yourself?

‍Hi, I'm Kate Strachnyi, the founder of DATAcated, providing brand amplification for companies focused on artificial intelligence (AI), machine learning (ML), and data science. I'm also an author, keynote speaker, and consultant with a passion for helping people succeed in the data space.

👀 What’s your day to day like in your current role as a founder of DATAcated?

‍As the founder of DATAcated, my days are filled with a mix of activities, from managing our community and engaging with our members, to creating content, developing media partnerships, and working on various projects (such as courses and presentations).

⭐️ What has been the biggest highlight of your career so far?

‍The biggest highlight of my career has been going off on my own and starting DATAcated. It was a big leap of faith, but it has been incredibly rewarding to see the community grow and to have the opportunity to help so many data professionals around the world. It's been an amazing journey, and I'm excited to see where it takes me next.

🚀 What's a data or AI trend you're watching this year?

‍Generative AI is the trend I'm watching closely this year. It's a rapidly growing area of AI that involves using algorithms to create new and original content, such as images, music, and text. With the help of deep learning models like GANs, generative AI has the ability to create high-quality content that is often indistinguishable from content created by humans.

💼 What advice would you give someone starting out in Data Analytics?

My advice would be to focus on building a strong foundation in the basics of data analytics, including statistics, programming, and data visualization. It's also important to stay curious, keep learning, and seek out opportunities to apply your skills in real-world scenarios. And of course, networking and building relationships with other data professionals is key to success in this field.

Try to grow your network at the same time as growing your skills - it will be helpful in landing a job or getting a promotion.

🤯 What’s the most important career lesson you wish you’d learned earlier?

One of the most important lessons I've learned in my career is the importance of protecting your time. As a busy professional, it's easy to get pulled in a million different directions and feel like you don't have enough time to get everything done. But the truth is, your time is one of your most valuable assets, and you need to be intentional about how you use it. It means being thoughtful about how you schedule your day, being selective about the meetings and projects you take on, and establishing clear boundaries with your colleagues and clients.

🧠 How would you learn Data Analytics if you had to start over?

If I had to start over, I would focus on building a strong foundation in statistics and programming, and then look for opportunities to apply those skills in real-world scenarios. I would also seek out mentors and network with other data professionals to learn from their experiences and insights.

I think getting projects completed and shared with your community as early as possible is very helpful in developing your skills while also growing your network.

🔥 Where can we find your amazing work?

You can find my work on my website, datacated.com, as well as on LinkedIn, where I regularly share articles, videos, and insights on all things data. I also post content on other social media platforms - YouTube, Medium, Twitter, Instagram, Facebook, and TikTok.

📚 What's your favorite book?

One of my favorite books is "Can't Hurt Me" by David Goggins. It's a powerful memoir that offers a lot of insight and inspiration for anyone looking to push past their limits and achieve their goals. What I love about this book is how raw and honest it is. David Goggins shares his life story in a way that is both inspiring and relatable, and he doesn't hold back on the challenges and struggles he faced along the way. He also offers a lot of practical advice and techniques for pushing past physical and mental barriers, and achieving your goals.

Another thing that sets this book apart is the interactive nature of it. The audiobook version of "Can't Hurt Me" includes additional commentary from David Goggins himself, and challenges and exercises to help you apply the lessons he shares to your own life. I highly recommend it!

🚀 Whenever you’re ready, there are 4 ways I can help you:

1. Book a coaching call with me if you need help in the following:

2. Promote your brand to 1,000+ subscribers in the data/tech space by sponsoring this newsletter.

3. Watch my YouTube videos where I talk about data science tips, programming, and my tech life (P.S. Don’t forget to like and subscribe 💜).

4. Follow me on LinkedIn and Twitter for more data science career insights, my mistakes and lessons learned from building a startup.

That's all for today

Thanks for reading. I hope you enjoyed today's issue. More than that, I hope it has helped you in some ways and brought you some peace of mind.

You can always write to me by simply replying to this newsletter and we can chat.

See you again next week.

- Admond