Inspiration

Our application development was inspired by the role of the assistant coach which is the true job of Machine Learning: Helping humans make better decisions and have better control over our lives. We looked up the biggest challenge that a coach faces in today's dynamic world of football. We unanimously came to the conclusion that finding a replacement to a retiring player or building good bench strength is the biggest headache of a coach. Every tom, dick and harry and their grannies have an opinion regarding who will be the best replacement for a player. Hence, we thought let's give machine learning a chance at that.

What it does

In the startup screen, the user is asked to pick-up the team that he/she is interested in. Then, the user is required to pick the player for which he/she wants the replacement suggestions. When the player name is selected the system shows the players' stats along with stats of recommended players so that user can review the differences in their performance.

How I built it

From the given dataset, we created metrics that are considered fundamental attributes for a player's performance in python. These metrics were then saved to a CSV file for quick loading of the solution. Then we applied the clustering algorithm for finding the closest match to a given player in R. The results were then displayed using shiny library of R.

Challenges I ran into

The biggest challenge that we faced was the extraction of complicated statistics from the data. We had to use our judgement call and take a chance to calculate the performance. Secondly, few terms were not clearly explained in the appendix file which resulted in lots of confusion

Accomplishments that I'm proud of

We as students are proud of being able to apply our skills to a practical life problem which has an impact on millions of lives. We are also proud of using different tools to create one complex product which seems very simple from the outside.

What I learned

I learnt about basic principles on how to handle a highly complex and large dataset. We also learnt different ways to divide our work as a team. Owing to our experience at US-SOCCER-HACKATHON we are more confident in working in a highly dynamic, goal oriented team environment. We would like to take thi opportunity to thank the organiser for hosting this awesome event

What's next for US-SOCCER-MLS-PLAYER-assistantCoach

We tried to create few modules like penalty shot predictor, player-specific man to man marking zones, opponent specific team formation recommender. However, we were not able to finish them due to lack of time. Hence, if given time and access to data we would like to finish these functionalities and create a true assistant coach.