Deck Clusterer

A python script for converting Hearthstone deck codes from a csv to declarations of their archetypes in csv format.

Requirements

Language that must be installed

• Python 3.8+

Packages that can must be installed
This can be done by following instructions in Setup

• python-hearthstone
• MatPlotLib
• Scikit-learn
• python-hearthstone-data
• PySimpleGui
• Tensorflow
• scikit-learn
• pandas
• Python Imaging Library
• backports.csv
• Requests
• BeautifulSoup4

Setup

Clone this repository first, then clone the hs-card-tiles repo https://github.com/HearthSim/hs-card-tiles within the Hearthstone-Deck-Clusterer directory.

git clone https://github.com/HearthSim/hs-card-tiles

Then properly install a Python version of atleast 3.8
Then at the first install, and after every Hearthstone Patch, Card Change, Or addition of Hero Portrait/Cardback
run setup.py found in the setup directory.

cd setup   
python setup.py    

This will update the card database, hearthstone library, and pull the most recent card tiles.

USAGE

To run the Clustering or Classification Program

python main.py

and follow instructions on the GUI

To Generate CSVs from Battlefy Data:
STARTDATE format: "YYYY-MM-DD" Note, 24 hour time so enter the Monday for a full weekend
ENDDATE format: "YYYY-MM-DD-YYYY" Note UCT Timezone so enter the Monday for a full weekend

cd CSVs/   
python getCSVS.py STARTDATE ENDDATE   

CSV formatting

Input

The first line in the CSV file will be the schema for the rest of the file. It will be a comma separated list of arguments with the same length as the rest of the rows of the file. Leave unused fields blank, and use "K" for keys (which will group the decklists and control what the image file is named as), and "D" for deck codes to be parsed by the script.

For example, a schema of K,D,D,D,D in the first line of the csv will indicate that the followings lines have the form "Name/Key, Deck Code #1, Deck Code #2, Deck Code #3, Deck Code #4", which a schema of "K,,D" will indicate the form of "Name/Key, [Irrelevant], Deck Code" and the program will group deck codes corresponding to the same key to the same deck image.

If the schema is not specified, a schema of K,D,D,D,... will be assumed.

Example input:

K,D,D,D,D
Akron,AAECAf0GBO0Fws4Cl9MCzfQCDYoB8gX7BrYH4Qf7B40I58sC8dAC/dACiNIC2OUC6uYCAA==,AAECAaoICCCZAvPCAsLOAqvnAvbsAqfuAs30Agu9AdMB2QfwB7EIkcECrMICm8sClugClO8CsPACAA==,AAECAQcC08MCn9MCDkuRBv8HsgibwgK+wwLKwwLJxwKbywLMzQLP5wKq7AKb8wLF8wIA,AAECAZICAv4BmdMCDkBf/QL3A+YFxAaFCOQIoM0Ch84CmNICntIChOYC1+8CAA==

also accepts full deck codes.

Output

For all modes
Outport will be of form: K,D,D,D,D,...
followed by the team name and deck calssification one each line

Example output:

K,D,D,D,D`   
Akron, Highlander mage, OTK dh, Enrage warrior, Zoo warlock   

Clustering Portion also outputs a csv of Labels that can be used for later classification usages into outputs/labels/NEW_labels/ this should be renamed for easier use.