Table of Contents

Blog Examples
Instructions
Repository Structure

MechWarrior Online Match Score Data Project

Disclaimer: This project is not affiliated with PGI or MechWarrior.

Project Goal:

This project aims to construct a dataset based on game results from MechWarrior online. Match results will be combined with mech weights to facilitate analysis of tonnage deltas between teams.

The resulting dataset will contain nearly 600 images converted to pipe delimited text files and mech tonnage by variant in the same format.

The combination of these datasets will allow statistical inference into matchmaking engine priorities.

Blog Examples

Please see [this link](link to blog) for a discussion on the use and results of these code files.

To run the scripts below begin in the Python sub folder.

• Whole screenshot: Code for sending the entire test image to the AWS Rekognition API. This is a test of the return and is not optimal for dataframe construction.
  python whole_screenshot_example.py

• Horizontal slicing: Code for sending horizontal slices of the test image to the AWS Rekognition APi. This example includes both a single row return as well as an example on assembling a dataframe from multiple API calls. This method optimizes cost at the expense of data cleaning time. python single_row_example.py

• Horizontal and vertical slicing combination: This example slices an image into individual cells and sends each cell individually to the Rekognition API. This method optimizes for data cleaning time at the expense of cost. python split_screenshot_example.py

• Dataframe output from OCR contains text files of dataframes produced by the example code files

• Images of dataframe output from OCR contains images of the dataframes produced by the example code files
  Analysis powered by kite

Instructions

• This project relies on Python 3.7
• Install requirements
• Set up your AWS credentials
• running web scrape code from the Python subfolder will generate a list of mech variants and tonnages: python run_mech_scrape.py
• running code for single screenshot: TBD
• running code for many screenshots: TBD

Repository Structure

• image data
  • data source (screencapture)
• Web Scraping: Uses the mech_scrape class to gather tonnage for each mech variant in MechWarrior Online.
• test data contains image files sliced from screenshots. Some images have been resized and/or threshed in greyscale.
• image data contains the test image used in the examples. A full set of screenshots is not currently available.
• image slicer converts screenshots to dataframes which can be saved as text. Two methods, horizontal, and a combination of horizontal and vertical slicing are available. The combination of horizontal and vertical slicing yields better results but requires more time and costs more in API calls.
• initial dataframes contains the initial dataframe output from screenshot conversions.
• cleaned dataframes contains dataframes that have been cleaned of errors created during the OCR process.