Emma McFall posted an update — Nov 30, 2021 05:17 PM EST

• Introduction: Existing paper’s objectives: forecasting college football game outcomes for approximately 130 teams in the 2019 season using various modeling techniques (including neural networks) This type of project is prediction based, specifically predicting wins college football teams would have in the 2019 season based on past win statistics. The paper we are implementing used a large variety of methods (including neural networks, random forests, k-nearest neighbors, stochastic gradient boosting, and a Bayesian regression model), and also predicted individual game wins, but we will use neural networks and look into the inclusion of other methods should time allow.
• Challenges: Pre-processing data because some types of data are missing for certain teams (for example smaller teams don’t track certain types of data so combining the data for each team can be a bit of a hassle) Deciding how to approach data cleaning, as there are many different statistics we can choose between. Also deciding how to set up our labels, because which team it is specifically matters
• Insights: Are there any concrete results you can show at this point? There are no concrete results we can show at this point as we are in a phase of preprocessing/coding the model
• Plan: Are you on track with your project?
• What do you need to dedicate more time to? Cleaning and uploading the data has proven to be harder than we expected. It is the hardest part of our project, and we need to dedicate more time to making sure it runs smoothly
• What are you thinking of changing, if anything? We considered changing what our model predicted to make it less team specific. Instead of predicting the winner for matchups, we considered predicting the win percentage for a season. We are sticking with our current model for now.

Emma McFall started this project — Nov 17, 2021 04:57 PM EST

Leave feedback in the comments!