18.6 Exercises

i2ds: Exercises

18.6.1 Predicting game outcomes in sports

A simple model for predicting possible outcomes of soccer games needs to assign probabilities to possible outcomes. For instance, when playing a soccer game, the outcome is determined by the number of goals scored by each team. If we can capture the probabilities of scoring goals, we can predict the probability of specific outcomes.

Study the following posts at the Learning Machines blog:

Tasks to address:

  1. Find some relevant data (ideally of an entire season) and tidy it. What are key variables that allow predicting game performance?

  2. Creating a model: Build a model that predicts the outcomes of games.

  3. Validation: Predict some outcomes in your data, and compare them to the actual outcomes.

  4. Out of sample prediction: Predict the outcomes for the next season and comapre them to the actual outcomes.

  5. Transfer to other tasks and domains:

    • What would change when we wanted to predict the outcome in another type of sport (e.g., boxing, basketball)?
    • What would change when we wanted to predict the outcome of political elections?