CP640 Machine Learning project (Spring 2022)

An independently researched and implemented machine learning project in the Master of Computer Science program at Wilfrid Laurier University.

Overview

I took historical Major League Baseball data from a Kaggle dataset (as determined by the project requirements) and ran some experiments in an attempt to predict future batting success.

Kaggle dataset source: https://www.kaggle.com/datasets/darinhawley/mlb-batting-stats-by-game-19012021 (external link)

The project, as submitted, was too large to be included on GitHub. As a result, the project .zip file (which includes the data files described in the Jupyter notebooks on GitHub) is available for download here (external link).

Project Deliverables

The complete set of project deliverables includes:

• a proposal document
• the Jupyter notebook(s) and source data (external link)
• a final report document, and
• a video presentation (external link)

======================

After Downloading the Project Zip

Pre-conditions:

• Local Jupyter Notebook server is installed and running

• harr2890_project.zip is extracted locally, with structure intact, where notebooks can be run

• Can create/write to data subfolder within the harr2890_project folder

Running harr2890_project Notebooks:

Please run the Notebook series in sequential, step order.

Step 1 (harr2890_project_step1_data_prep)

• General data preprocessing; must be run before all other steps

Step 2 (harr2890_project_step2_hof_data_prep)

• Hall of Fame Approach preprocessing; must be run before Step 3

Step 3 (harr2890_project_step3_hof_modelling)

• Hall of Fame Approach modelling (selection and evaluation)

Step 4 (harr2890_project_step4_ops_data_prep)

• OPS Approach preprocessing; must be run before Step 5

Step 5 (harr2890_project_step5_ops_modelling)

• OPS Approach modelling (selection and evaluation)