Hack Arizona Experience – Project

Hack Arizona:

This year’s Hack Arizona was organized January 12th to 14th. This hackathon is a yearly, weekend long event that was originally started at the University of Arizona in 2015, it has grown both in students and companies attending the event. The purpose of such hackathons is to provide people with a productive environment and the opportunity to do something great. There is no restriction on what people can work on and there is no minimum or maximum requirement, the event is simply in place to give guidance and support. There is also no skill requirement or advised skill level to participate, as the workshops and challenges range from absolute beginners to experts, giving everyone a fair chance at some prizes.

Event Proceedings
The event usually starts out with an opening ceremony, where first the organizers introduce the themselves and talk briefly about the event. This is followed by company representatives who discuss their challenge for the year (and often give out valuable prices to lucky individuals). These company challenges can be as simple as solving a puzzle, or as convoluted as creating a fully operational VR game. The companies also give out pre-established prices for these challenges, usually according to the difficulty of said challenge. If someone were to miss the opening ceremony, they can often find this information online or can ask the companies personally at their booth during the event. 

There is usually a set time (9 pm this year) where the hacking may officially begin, this is when people can start working on the challenges and compete for the various awards of the hackathon. While the event is in progress, participants are provided with free meals for breakfast, lunch, and dinner; and with further free snacks and sodas throughout the day. Throughout the hackathon, there are schedules presentations and lectures, which participants can attend freely. At this point, people can do anything they want, and work anywhere they want, though it’s advised to work in the library grounds as those are all strictly restricted for participants during the weekend. 

The event lasts the entire weekend, however “hacking” ends Sunday early morning, at which point all challenges are completed and teams participating have to submit their work to applicable challenges and awards. Awarding procedures vary, however there is a large event hall (in this year’s case in the Student Union) where teams can then present their projects to companies which evaluate their work and score them accordingly. Afterwards, there is an award ceremony, where winners of the major challenges and hackathon awards are announced and awarded their prices. These prices vary from year to year, and range from shirts, to gadgets, and even up to a USD 5,000 cash price in some cases.

Personal Experience

With this said, here are my personal experiences and the project we have worked on within the confines of this year’s Hack Arizona. We have attended this year’s event with a small group of friends (which is a great idea if you are new to the event as it makes things much less intimidating), with initially no intentions on partaking in challenges. We were more concerned at that time with the sizeable portion of free gifts one can get from the event. 

The size of the event itself was mesmerizing as there were over 800 participants present!! We haven’t participated in the activities on Friday, but rather just enjoyed the atmosphere of the event. When we came back Saturday morning, we started off doing smaller challenges, such as writing a short program, or decrypting a password hash (the latter of which we got first place luckily). 

While doing this, we got inspired by the challenge of Paradox, a local company specializing in AI assisted hiring systems. The challenge was to utilize machine learning to help soothe the employment process for potential candidates, and thus help companies hire better employees. Given our interest in machine learning and the $5,000 cash price, we quickly got to work. To get some basic understanding of some of the pre-existing machine learning algorithms, we decided to attend a seminar on Artificial Intelligence and Machine Learning algorithms. Here we were introduced to the idea of a K-means clustering machine learning approach. This served as the foundation for our project later on. We set to work, and managed to create an algorithm that can potentially help a company evaluate personality fit based on trait ratings.

K-means clustering results (not ours, these are 2D)

The algorithm itself relies on a fairly simple idea. The company first surveys their pre-existing employees on any given number of characteristics, on Likert scales. For example, they could pose questions such as: “Do you prefer working alone or closely with your team?” Please keep in mind that it is very simple to compare people using regular methods if there is just one indicator, but it gets exponentially harder as we factor in more and more values. 

This is where the K-mean algorithm comes in very handy. The goal of this algorithm is to create clusters of data points. Each data-point equals an employee, with the characteristics being the coordinates of the point. This is easy to imagine in 3 dimensions, but it is equally workable in even 100 dimensions. Machine learning is used to identify the central points in the clusters, and then move them appropriately and effectively as more and more data points are added. The final result of our algorithm basically tells the employer whether the new employee is a good fit in either pre-existing cluster in the company, and if so, which team would he work best with. (A 3D representation of a small sample data set is shown in Figure 2.) Please note that the program can be easily adjusted to company needs, by changing cluster sizes, giving different characteristics different importance, or changing the dimensions of the clusters.

Sample 3D clusters with new data (red)

We finished our project by Sunday around sunrise, and submitted it for as many applicable categories as possible (even some that were just a little far-fetched). We have quickly compiled a presentation on the basics of our algorithm and what we used to create and compile the code. We presented our project Sunday, just before noon, to several teams and judges. At this point we had absolutely no idea what other teams did, and we were rather intimidated by the unknown competition. 

While we didn’t expect to take come any of the prizes, we ended up winning three major categories! More specifically, we won: 

• the Best New Hack category, 
• the Machine Learning category, and 
• Amazon’s Best Use of Data Intelligence category. 

All in all, it was beyond an amazing experience, through which all of us have learned a lot, not just about machine learning, but about the value of dedication and hard work. It’s surprising how much work one can accomplish in such a short amount of time with the right motivation and productivity.

More info
Hack Arizona Website: http://hackarizona.org/
A K-means explanation: https://mubaris.com/2017/10/01/kmeans-clustering-in-python/

Questions

1. What do you think of hackathons, have you ever wanted to or attended one? 
2. What machine learning ideas can you think of to enhance the employee hiring process? 
3. If you were given the chance and a team to build any project on a weekend, what would it be?

If you have any questions related to my post, or would like to add anything that you think I have missed about the hackathon, please don't hesitate to comment!