Inspiration
Workspace allocation is becoming an increasingly important problem for modern businesses. Rising land costs are decreasing the availability of office spaces, and the increasing reliance on collaboration by teams in contemporary businesses means that it is becoming more and more crucial to place specific groups of employees near each other in an optimum way. At the core of our project is a machine-learning algorithm that makes solving this problem a bit easier.
What it does
Our algorithm takes in information about an office space and all the teams of employees within it. It then uses the information to allocate the teams of employees within all the floors of the office space in such a way that all the teams are as close as possible to the teams they prefer being near to. At the same time, the algorithm also takes into consideration the space constraints of the building, so that none of the floors get too crammed. Our program then visualizes that information in a friendly and human-readable manner.
How we built it
The front-end of the website that takes in information about the office space was made using React. The algorithm to optimize the space allocation using the user-defined information was written using Python. The data visualization was performed using C# and Unity, and compiled into a WebGL project that can run on a React frontend.
Challenges we ran into
We had difficulties transferring the data from the frontend into the C# data visualization script. We also faced some challenges quantifying the concept of compatibility between two teams. Extending the compatibility between two teams to factor in multiple teams was an ever bigger hurdle that we had to cross.
Accomplishments that we're proud of
We were successfully able to create a proprietary algorithm for the prompt that was completely different from anything available and tailor-made to this specific problem. We were also able to feed the data into a Unity script that generated intuitive and beautiful visualizations from that information. Finally, the React frontend makes it really easy for users to enter all the information needed by the algorithm in an understandable manner.
What we learned
Solving this problem was a well-learned lesson in optimization and compartmentalizing a big problem into smaller, more manageable ones. We also learned to apply a lot of the machine learning and data structures lessons from school to real-world situations. Finally, making a project that uses 3 different libraries and programming languages made us appreciate the intricacies behind transferring data between different frameworks and working with it in an efficient manner.
What's next for Who's Where?
Our algorithm has a lot of room for improvement. In the future, we can take more factors into consideration, such as seating preferences for different employees (sitting/standing desks, the importance of sunlight in their spaces, etc. ), the utility of specific teams to the company, and many qualitative pieces of data. We can also improve the data visualization to represent the exact geographical arrangement of teams within a single floor.
Log in or sign up for Devpost to join the conversation.