This project is still under development. I hope to invest more time into this project as school workload permits in the coming weeks. Some future plans/visions include:
- Continue exploratory data analysis (
eda.ipynb) to identify potential features for predictive modeling - Revisit the scrimmage database to engineer new features
This repository contains a set of Python scripts designed for cleaning and analyzing Overwatch match data. The project structures the match data to identify key gameplay events such as staggers and first kills within each fight, enriches the data with hero role information, and provides tools for connecting to and querying a database. The analysis part of the project helps in understanding team performance by fights won.
cleaning.py: Contains helper functions for cleaning and structuring Overwatch match data, including identifying "stagger" kills and first kills in fights, and mapping heroes to roles.collect_data.py: A script for connecting to a database to retrieve match data, clean it using functions fromcleaning.py, and prepare it for analysis.db_connect.py: Includes functions for establishing connections to a PostgreSQL database, and fetching data via PostgreSQL queries.hero_mappings.py: Defines mappings between Overwatch heroes and their respective roles (Damage, Tank, Support), used for enriching the match data.two_way_table.py: A simple analysis script that groups cleaned data by attacker team and whether they won the fight, summarizing team performance.
- Python 3.8+
- pandas
- psycopg2
- python-dotenv
- Clone the repository to your local machine.
- Ensure you have Python 3.8 or newer installed.
- Install the required Python packages:
pip install pandas psycopg2 python-dotenv - Create a
.envfile in the root of the project directory containing your database connection URL:prod_url=YOUR_DATABASE_CONNECTION_URL_HERE
-
Data Collection and Cleaning:
- Execute
collect_data.pyto collect match data from your database, clean it, and prepare it for analysis. This script integrates data cleaning, role enrichment, and database interactions seamlessly. - Ensure your database contains the required tables and schema as expected by the query in
collect_data.py.
- Execute
-
Data Analysis:
- After cleaning, you can use
two_way_table.pyto analyze team performance based on the cleaned data. This script provides a simple analysis, grouping data by attacker team and whether they won the fight. - I've provided a simple logistic regression model under
simple_logit.ipynbas a sample of potential work that can be done with the current data. The model is far from robust, but does confirm preexisting beliefs about the game.
- After cleaning, you can use
Contributions are welcome! If you have suggestions for improving the code, please open an issue or submit a pull request.
- Remember to update your
.envfile with the correct database connection URL before runningcollect_data.py. - The scripts assume specific database schema and table structures. Adjust the SQL queries in
collect_data.pyanddb_connect.pyas necessary to match your database setup. - For detailed analysis or additional functionality, consider extending
two_way_table.pyor creating new analysis scripts based on the cleaned data.
If you have any questions or need further assistance, feel free to reach out or open an issue in the repository.