Data-Match is the winning project of the Datathon FME 2024 AED Challenge, aimed at simplifying the process of team formation.

Our tool uses basic Machine Learning techniques to match participants based on their skills, preferences, and compatibility, and it is designed to ensure that everyone has an enjoyable and productive experience during the Datathon.

The platform includes three main components:

• User-Friendly Form: Collects relevant participant data efficiently.
• Simple Matching Algorithm: Groups participants into compatible teams.
• Chatbot Assistant: Provides insights into team recommendations and answers user questions.

• Create: .env file with a groq LLAMA_APY_KEY
• Execute: content_based.py
• Open: /web/index.html

Taking a look at the user database and applying Dimensionality Reduction techniques (PCA), we can observe various user clusters. We can see that the individuals that are recommended by our system are from the same cluster as the original user, allowing us to see that they are similar users.

Our tool also recommends groups of friends that have already applied together, where we can see that the user still matches with teammates from the same cluster.

We can also take a look at the Variance of the users that our system recommends. As we can see on the folowing chart, the variances are very low, indicating that the users are, indeed, very close to each other.

First, we process the data from the user database, selecting the most relevant variables based on match compatibility. Next, we standardize all the data to ensure that all values fall within the range of 0 to 1, balancing the weights across the variables.

Once the data is standardized, we group users into teams, taking into account that some users have applied with friends. We then apply the cosine similarity function to analyze the distances between users in a multidimensional space using the content based filtering algorithm. Users are ranked by proximity, and we select a group of four suitable members. Additionally, we create a backup pool of similar users who can be considered if the initial match is unsuccessful.

We’ve designed a modern, intuitive user interface with a strong focus on ease of use.

To start, we’ve developed a participant dashboard that showcases relevant information, such as upcoming events, ensuring users have quick and easy access to key details.

The registration form has been carefully structured to gather a large amount of relevant data while minimizing complexity. The form is divided into sections, making it easier and less time-consuming to complete.

Once registered, our ML matching system analyzes the data to identify the best team combinations, presenting the results to the user.

Additionally, an AI-Agent powered chatbot is available to answer detailed questions about both individual participants and the team as a whole, ensuring transparency and explainability in our recommendations.

Users can save, swap, and adjust team members until they are satisfied.

Once the perfect team is formed, joining is just a click away.

Team building has never been simpler! ☀️