Overview
- Data science hackathons are a great way to test, improve and build your data science skillset
- Hear from top data science experts like SRK, Dipanjan Sarkar, Rohan Rao, and more in these full session videos!
Introduction
I love the multi-faceted nature of data science. It’s an amalgamation of our coding skills, statistical knowledge, business knowledge, and other various skillsets.
If you’re a budding data scientist, the question is – How do I improve my skills and how do I reflect these skills in my professional career? How do I test and improve my theoretical knowledge in a practical manner?
The answer to all of this – Data Science hackathons!
A lot of organizations are organizing their own data science hackathons and competitions, often as part of the hiring process. These competitions help you gain the attention of potential recruiters, fellow data scientists, and help you gain practical know-how of the dimensions of machine learning! So are you ready to get started on your own data science hackathon journey?
But here’s the caveat – hackathons can be overwhelming for beginners. How do you start? Where do you start? What type of problems should you start solving? etc. Even intermediate performers struggle to maintain a good rank on the hackathon leaderboard.
So what is the secret behind some of the top performers in these competitions?
We’ll get to know the answers from the data science hackers themselves! Top data scientists like Sudalai Rajkumar (SRK), Rohan Rao, Kiran R, Pavel Pleskov, and Dipanjan Sarkar share their journey and their hacks, tips, and tricks so let us jump right in!
These excerpts have been taken from Analytics Vidhya’s free course:
I would highly recommend you to go through these talks, it will provide you with invaluable insights into the world of data science and hackathons.
Talk #1 – Effective Feature Engineering – A Structured Approach to Building Better ML Models by Dipanjan Sarkar
“Coming up with features is difficult, time-consuming, requires expert knowledge. ‘Applied machine learning’ is basically feature engineering.”
– Prof. Andrew Ng
If you take up any data science project or competition, 70% of your time is generally consumed in the phase of data preparation. This includes feature engineering, data wrangling, and visualization. This step can make all the difference in your project.
In this talk, Dipanjan Sarkar expresses feature engineering as an art which may require constant attention to detail and business domain knowledge rather than building a bunch of random features.
Even with the advent of automated machine learning tools (AutoML) and automated feature engineering libraries, creative and innovative hand-crafted features can often be the deciding factor in winning competitions or getting a real-world project from being a proof of concept to being deployed in production.
This session is more of a practical demonstration than talk! Dipanjan fires up his notebook and illustrates feature engineering by solving two case studies:
- Predicting taxi fare prices around New York City. This is a regression problem at a scale where you will be dealing with millions of data points!
- Dataset pertaining to E-commerce product reviews and ratings. Here, we tackle a classification problem of trying to predict recommended product ratings based on the customer review descriptions – a classic NLP problem!
Key Takeaways from this session
- Understand the need and importance of feature engineering, data wrangling, and visualization
- Look at real-world problems where hand-crafted features and human domain expertise might be more useful than randomly building a bunch of features or models using automated techniques or meta-heuristics
- Hands-on demonstration of feature engineering and machine learning modeling on big datasets having millions of data points
- Comprehensive coverage of popular feature engineering techniques for structured data including numeric, categorical, temporal, and geospatial features
- Comprehensive coverage of popular feature engineering techniques for unstructured text data including BoW, TF-IDF, and even newer deep learning models (word2vec, FastText)
You can refer to the following article to read more:
Talk #2 – Automating the Machine Learning Pipeline with AutoML by Dr. Sunil Kumar Chinnamgari
AutoML can be a very controversial word/technology for a data scientist. Many consider it as a threat to their current role but Dr. Sunil plans on using this technology in his own favor.
AutoML is a suite of packages that are provided by various open-source contributors. These AutoML packages are capable of auto-selection of machine learning models and optimizing hyperparameters which takes a high amount of time as a manual job. Now the core question is – How does this help in your next data science hackathon?
Another great advantage of this technology is that it can be used to create benchmarks which essentially means that you get to know what level of performance is possible for a particular problem statement. Dr. Sunil displays some great results by applying AutoML packages to Analytics Vidhya’s Loan prediction dataset using Python.
Some of the AutoML packages used are –
- AutoWeka
- Autosklearn
- H2O AutoML
- TPOT (Tree-Based Pipeline Optimization Tool)
Key Takeaways from this session
- Get acquainted to the world of AutoML
- Is AutoML going to take away the job of a data scientist or is it a complementing technology?
- What are the different packages for AutoML?
- Learn how to solve different problem statements using AutoML tools
- Comparison of various AutoML tools through their performance measurements on several problems
Talk #3 – What Sets the Top Hackers Apart? – A Panel Discussion of Top Data Science Hackers
Hackathons and coding challenges are a great way to showcase your skills as a data science professional. But what on earth will it take to repeatedly perform well in these intense data science competitions? Well, this panel discussion might give you the answers!
The insightful discussion amongst India’s top data scientists – Kiran R, Sudalai Rajkumar (SRK), Rohan Rao, Sourabh Jha, Sahil Verma, Mohsin Hasan Khan is structured into 3 categories:
- How did they get started with the hackathon journey?
- What are some of the tips and tricks that these hackathons stars have used?
- How do hackathons help you in your professional life, be it academic or work-related?
“I had no exposure to coding, and on the first day [of the job] they gave me a Linux machine and I was searching for icons how to write Python code for 3-4 hours and that’s how I started my journey, but slowly I started picking up.” – SRK on how he started his journey.
Other experts have a similar story so don’t hesitate in starting out your journey in hackathons. 🙂
The discussion also features the most coveted question – Python or R? While the majority of hackers shifted to Python, Rohan Rao still loves R. Moreover, having knowledge of both languages is a plus.
If you want to become a top data science competition performer than this talk is a must-watch!
Key Takeaways from this session
- Top Hacks and advice for data scientists competing in hackathons
- What is your approach in the beginning when you see a problem statement in a hackathon?
- How important is computing power when it comes to winning hackathons? What tools do you use?
- Is competition/hackathon work useful in real-world production environments?
Talk #4 – Top Hacks from a Kaggle Grandmaster by Pavel Pleskov
Data Science competitions are prestigious – they help to increase the learning curve (and showing off to your peers). But winning data science hackathons continuously is a tough task! Just think about the number of obstacles in your way:
- A brand new problem statement we haven’t worked on before
- A plethora of top data scientists competing to rise up the leaderboard
- Time crunch! We have to understand the problem statement, put together a framework, clean the data, explore it, and build the model in a matter of a few hours
- And then repeat the process!
A single decimal point could be the difference between the top 10 and the top 50. Isn’t this why we love hackathons in the first place? The thrill of seeing our hard work pay off with a rise in the leaderboard rankings is unparalleled.
Pavel Pleskov, former Kaggle Grandmaster, shares some of his secret tips, tricks, and hacks which will help you to achieve outstanding results in any Data Science competition.
Key Takeaways from this session
- Learn some of the great tips and tricks by a former Kaggle Grandmaster
- Learn about blending ROC-AUC metrics, tweaking, reproducibility, and much more!
- How to continuously outperform in all the hackathons
Talk #5 – Feature Engineering for Image Data by Pulkit Sharma and Aishwarya Singh
Well, we have settled on a common ground for tabular data – feature engineering is one of the most crucial steps in a machine learning pipeline which can make or break your model. The case remains the same for image data as well. But wait – since we use deep learning here, why is feature engineering important here?
There’s a strong belief that when it comes to working with unstructured data, especially image data, deep learning models are the way forward. Deep learning techniques undoubtedly perform extremely well, but is that the only way to work with images? Not all of us have unlimited resources like the big technology behemoths such as Google and Facebook. So how can we work with image data if not through the lens of deep learning?
We can leverage the power of machine learning! That’s right – we can use simple machine learning models such as logistic regression, support vector machines (SVM), or decision trees. In case, if these machine learning algorithms are provided with the right data and features, they can perform adequately and can even be used as a benchmark solution.
Key Takeaways from this session:
- Learn how to extract primary features from images, like edge features, HOG and SIFT features
- Extracting image features using Convolutional Neural Networks (CNNs)
- Building Image classification model using Machine Learning
- Performance comparison among primary and CNN features using Machine Learning Models
You can refer to the following articles to read more:
- Feature Engineering for Images: A Valuable Introduction to the HOG Feature Descriptor
-
Image Augmentation for Deep Learning using PyTorch – Feature Engineering for Images
End Notes
To conclude, in this article, we discussed some of the talks that will help you get started in your future data science hackathons and hopefully you rope in a double-digit rank. These talks traversed through rockstar data scientists like Sudalai Rajkumar (SRK), Pavel Pleskov, Rohan Rao, and many more.
There is a lot of difference in the data science we learn in courses and self-practice and the one we work in the industry. I’d recommend you to go through the following crystal clear free courses to understand everything about analytics, machine learning, and artificial intelligence:
- Introduction to AI/ML Free Course | Mobile app
- Introduction to AI/ML for Business Leaders Mobile app
- Introduction to Business Analytics Free Course | Mobile app
Let me know your thoughts on hackathons in general and how to become better at data science competitions. Happy Learning!
