SoccerAssistant

Introduction

SoccerAssistant is a two-system chatbot we created for the Northwestern men's soccer team. It is designed to enhance performance both on and off the field. The first system focuses on competitive strategy, helping coaches and analysts review opponent tactics, assess strengths and weaknesses, and recommend strategies for upcoming matches. The second system is dedicated to analytics, providing detailed insights into player and team performance, aiding in player evaluation by tracking key metrics, and helping with team evaluation to identify areas for improvement. Together, these systems support the team’s decision-making process, helping the coaching staff make data-driven adjustments to improve both individual and team performance.

Table of Contents

1. Introduction
2. Key Features
3. Setup Instructions
   • Clone the Repository
   • Set Up Virtual Environment
   • Set Up Environment Variables
   • Run the Application
4. Agentic Workflow
5. Contributing

Key Features

• AI-Powered Chatbot Interface seamlessly integrates with a dynamic 4-agent backend for real-time data processing.
• Instant Summary Analytics to generate actionable insights, player comparisons, and interactive visualizations.
• Opponent Scouting Intelligence leverages predictive analytics to identify weaknesses in opposing teams for strategic game planning.
• Advanced Player Performance Tracking, including metrics like goals, assists, passing accuracy, and heatmaps for in-depth analysis.
• Comprehensive Team Analytics, offering insights into possession rates, attacking efficiency, shots on target, and defensive resilience.
• Data-Driven Performance Optimization through pinpointed areas of improvement using AI-aggregated team statistics.

Setup Instructions

Follow these steps to set up and run the SoccerAssistant project:

1. Clone the Repository

git clone https://github.com/ayush9818/SoccerAssistant.git
cd SoccerAssistant

2. Set Up a Virtual Environment

For Linux/Mac:

python3 -m venv .venv
source .venv/bin/activate

Install the required dependencies:

pip install -r requirements.txt

3. Set Up Environment Variables

Create a .env file in the project root directory and add the following variables:

cp env_template .env

• OPENAI_API_KEY : Replace it with appropriate API Key

4. Run the Application

1. Running the ChatAgent

   export PYTHONPATH=$(pwd):PYTHONPATH
   python chat_agent/app.py

The app will be running on http://127.0.0.1:7860

2. Running the SoccerStrat ChatAgent

   cd strategy_agent/SoccerStrat
   streamlit run soccer_app.py

The app will be running on http://localhost:8501

Agentic Workflow

Agentic Workflow Overview

1. Soccer Assistant Bot:

   • The system represents a multi-agent architecture designed to process user queries in the context of soccer-related data.
   • It consists of multiple specialized agents, each focused on a specific type of information:
     • Team Agent: Handles team-related information and statistics.
     • Player Agent: Focuses on player-specific statistics and performance.
     • Strategy Agent: Analyzes game strategies and tactics.
     • SQL Analytics Agent: Processes structured data queries using a Text-to-SQL capability.

2. Query Handling:

   • The user inputs a query into the system.
   • A central LLM-based Selector automatically determines which agents are relevant for the query. It intelligently re-routes the query to one or more agents.

3. Agent Processing:

   • Each agent processes the query independently based on its domain of expertise.
   • For example:
     • The Player Agent might fetch player statistics.
     • The SQL Analytics Agent might retrieve structured data from a database.
     • The Strategy Agent could analyze tactics or provide insights into game plans.

4. Response Summarization:

   • The outputs from all the agents are merged and summarized into a single cohesive response.
   • This ensures that the user receives a comprehensive answer to their query, drawing from all relevant domains.

This workflow ensures efficient and accurate handling of diverse soccer-related queries by leveraging the specialized capabilities of each agent. It uses LLM-driven intelligence for routing and summarization, creating an intelligent, user-friendly system.

Contributing

• Ayush Agarwal
• Kevin Li
• Jason Huang
• Omar Shatrat