ODI PRE MATCH OUTCOME PREDICTION AND STRATEGY RECOMMENDATION SYSTEM

What is it?

This is a pre-match outcome prediction system that takes in historical match data to predict win or loss. We model the game using a subset of match parameters, using a K- nearest-neighbour clustering algorithms and SVM classifier. It also suggests player performance along with their preferred roles in the match.

Requirements:

• Python >= 3.0
• Linux/Windows Operating System with atleast 1GB of RAM
• A browser to view the application

Dependencies:

cycler==0.10.0
Flask==0.10.1
itsdangerous==0.24
Jinja2==2.8
MarkupSafe==0.23
matplotlib==1.5.1
nltk==3.1
numpy==1.10.4
PyMySQL==0.7.2
pyparsing==2.1.0
python-dateutil==2.4.2
pytz==2015.7
scikit-learn==0.17
scipy==0.17.0
six==1.10.0
Werkzeug==0.11.4
wheel==0.24.0

Setting Up:

• Install Python

• Install Pip

• Copy code into a suitable project directory

• Install virtualenv: $ [sudo] pip3 install virtualenv

• Setup virtualenv: $ cd /path/to/project/ && virtualenv env && source env/bin/activate

• Install dependencies: $ pip3 install -r requirements.txt

Running the script:

Download score cards from Howstat and store it in a directory named dataset/scorecard.
Dump the structured data onto the database.
Modify paths under code/core/config.py to the system path.

1. Data preprocessing:

$ python3 core/preprocessing/scorecard/extract.py

2. Run web application:

$ python3 app.py

Open browser and visit http://127.0.0.1:5000/