How to win a hackathon: Two days. Twelve hours of intense coding. More than 40K data points and 700 lines of code. Tons of coffee.
Tecknoworks wins first place.
This pretty much sums up our experience at the Microsoft Azure Machine Learning Hackathon. For me personally, it has been an incredible learning experience that I would recommend as the top item on each coder’s bucket list. Based on it, I would like to share some key tips and tricks to make the most out of a hackathon experience and have a real chance of winning it. It applies to hackathons in general, with some particularities on Machine Learning hackathons.
First of all, the benefits of participating in a hackathon are huge and can represent a giant leap ahead, changing a lot about one’s beliefs and knowledge on a topic. For me, the biggest takeaway was realising that Machine Learning is the future of computing and science. Being able to predict the future with computational power was always a dream of mankind since the early days of machines, and we are now lucky to live in a time when this is made so easy through machine learning platforms. The power of knowing things before they happen offers so much control that it will surely boost businesses in the future.
To sum up, here are my top 5 reasons why I would recommend you attend a hackathon:
By now, you might be convinced that attending a hackathon is what you want to do next. So how to make the most out of the experience and have a real chance at winning? Here is my advice:
First and most important step would be to be organised and prepared. Always take time and book some workshops with the team before the main competition. Try to predict what your needs will be and plan carefully. We spent a lot of time in mastering R on-site rather than focusing on the actual implementation. Additionally, try to get everyone aligned to what you are trying to achieve and try out some brainstorming sessions to get the most out of people. There are different techniques out there for this so please feel free to exercise.
Choose a team leader. Be it Machine Learning or any other topic, having a subject matter expert on the team gets you a long way ahead. Make sure the team leader knows the team and the project you are trying to implement and can delegate tasks to the appropriate person.
Assemble a strong and varied team. Having a mix of skills in the team and making it fit together is a game changer. Don’t try to do everything yourself and delegate to the team member that has the knowledge to bring that task to a successful end, as time is not on your side. Our team was composed of a Business Analyst, 2 BI experts and one data scientist that completed each other wonderfully.
Chose a real life problem with an impactful and original solution. This can be a great source of motivation and can weigh a lot in the jury’s decision. The problems that were tackled at the hackathon were very diverse, ranging from classical insurance fraud detection solutions to the more extraordinary solutions, such as bird song classification based on wave files. We chose two real life challenges based on the experiences of one of our clients in the insurance industry related to predicting the number of patients on a certain drug cluster and determining their loyalty. We managed to predict the outcomes with a high forecast accuracy and the solution proved its potential use by pharmacies and insurance companies worldwide, thus it was a winner.
Take time to do pair-programming. Get yourself out of that chair and stay behind another team member and try to tackle a problem together. It gets you way ahead in implementing something correct and fast, from the very start.
Get some rest— before, during and after the hackathon. Your mind is your greatest asset – if it does not function at an optimum level you will get the feeling that everyone else is far ahead and you cannot make progress, which can discourage you. Giving yourself short pauses to rest is much better than continuously working and pumping caffeine and energisers.
Get your data straight. If you have a data problem, as we did for the ML hackathon, then you definitely need to spend time in modelling the data well before the competitions starts. Otherwise, the pressure of not seeing good results due to time spent in identifying trends will get your spirits down.
Mingle with the attendees. Chat with other participants that might have the same idea and work towards getting help or sharing your experiences. It helps you grow your network and might provide new opportunities in the future.
Boost up your confidence when it comes to presenting the idea. I can give you a strong inside tip here — “selling” your idea well already gets you halfway to the winners side. Let team members with selling skills practice their speech and the way they will present the solution. Tell a story and be one with the audience.
Focus and prepare the environment before the hacking starts. I agree that hackathons are also ways or learning and acquiring new skills but at least take the time and test things beforehand and don’t affect the team’s chances of winning by getting up to speed with the tools you need to use on-site.
Visuals are key. These events are usually aimed at developers but all in all, we are visual beings— seeing things helps in making decisions. Focus on showcasing something that the audience can relate to and does not require time in understanding.
So there you have it — proven tips on how to win a hackathon. Let us know how your next one goes!