Avaya A's Senior Project Blog
 |
Project Title: Exploring AI’s Predictive Power: A Study of Sports Outcomes BASIS Advisor: Sivashanmaugapriya Sankaralingam Internship Location: Honeywell Onsite Mentor: Latif Bansal, Computer Engineer |
Project Abstract
Artificial intelligence holds immense potential to reshape industries and improve society, particularly through its ability to predict future events. Through this project, I will investigate the extent of AI's predictive capabilities by focusing on sports outcomes. Not only will it be able to revolutionize the sports industry—enabling innovative playing strategies, enhancing fan experiences, and potentially creating new sports— but these advancements could extend to broader applications, such as disaster forecasting or medical predictions. This research contributes to the growing field of AI and computer science, aiming to uncover how far our current technology has progressed in forecasting complex systems. To research AI’s predictive capabilities, I will use Meta Chameleon, an advanced AI platform, and provide it with diverse sports-oriented datasets spanning multiple years. These datasets will include variables such as player attributes (strength, agility, consistency, health conditions), team health, and match histories. By analyzing the AI's predictions against actual outcomes, I will measure its accuracy and delve into the reasoning behind its conclusions. Along with this analysis, I will also identify patterns and factors behind AI’s decision-making. While AI may not achieve precise accuracy, its predictions will reveal trends that highlight its growth potential. These results could demonstrate the future viability of AI in forecasting not only sports outcomes but also critical events in other fields, paving the way for innovative applications and further advancements in AI technology.
Week 10: The Final Prediction
Hello everyone! Ten weeks ago, I set up Meta Chameleon to answer what seemed to be a simple question: Can artificial intelligence accurately predict the outcomes of football games? After weeks of experimenting, training, refining, and testing, it was time to put the model to the test. I selected ten recent NFL games, fed the... Read More
Week 9: Challenges and Reflections
Hello everyone! After eight weeks of working on Meta Chameleon, from setting up models and datasets to making predictions and drawing connections to the real world, it feels surreal to say that I’m nearing the final stretch. This week, I wanted to take a breath, reflect on the entire process, and share some of the... Read More
Week 8: How AI Forecasting Can Save Lives
Hello everyone! Up until now, I’ve been focused on sports - training Meta Chameleon to analyze patterns in stats, make predictions about game outcomes, and improve its accuracy week by week. But this week, I took a step back and asked a deeper question: If AI can predict something as chaotic as sports, what else... Read More
Week 7: Feature Selection – Can Better Data Improve AI Predictions?
Hello everyone! Last week, I evaluated Meta Chameleon’s accuracy in predicting sports outcomes. While it performed better than simple models, there was still room for improvement. I realized that choosing the right data features is just as important as the AI model itself. This week, I focused on feature selection - deciding which pieces of... Read More
Week 6: Compiling Information
Hello everyone! This week, I worked on compiling information for my datasets. After setting up Meta Chameleon last week, I was excited to see how well it could predict sports outcomes. But before trusting its results, I needed to answer a key question: how do we measure the accuracy of AI sports predictions? In this... Read More
Week 5: Setting Up Meta Chameleon
Hello everyone! This week, I worked on setting up Meta Chameleon, because, unlike other machine learning models, this model has to be set up in a certain way through a GitHub repository that looks like this: Once I read through all of the directions, I started by installing the required libraries, like plotly, numpy, pandas,... Read More
Week 4: Could Robots Replace Sports Announcers?
Hello everyone! This week, I wanted to delve into how AI might have a role other than simply being able to predict sports outcomes. I looked into if AI, and more specifically robots, could potentially replace sports announcers! With advancements in natural language processing and real-time data analysis, AI can create instant commentary, provide in-depth... Read More
Week 3: Hypothesis, Framework, and Research Design
Hello everyone! This week, I worked on finding a hypothesis and developing the aim of my research. The hypothesis I’ve decided on is: If Meta Chameleon is given prompts to predict current team sports using sports-oriented datasets, then it will respond with substantial levels of accuracy (~70%). The AI will achieve this accuracy by selecting... Read More
Week 2: The Role of Data in AI Sports Predictions
Hello everyone! Today, I want to dive into a major part of my project - how data plays a massive role in AI sports predictions. Data powers the AI engine, and the better the data is, the more accurate the predictions would be. I’ll be providing Meta Chameleon with a set of data points to... Read More
Week 1: An Introduction & An Inside Look at Meta Chameleon
Hello everyone! I’m excited to share my journey with you as I explore the endless possibilities of artificial intelligence and, more specifically, its potential to predict sports outcomes. I hope to use this blog as a space where I can show all of my progress and any challenges that arise while working on my senior... Read More