Cricket-Series-Win-Predictor-Using-Machine-Learning

Predicted outcomes of future cricket series using Machine Learning tools

Overview

In this project, I aim to predict the team which wins future cricket series based upon historical win ratio between various teams. This project was basically completed utilizing data taken from Kaggle and ESPNcricinfo.

Tools Used

1. Pandas - Mainly used for data manipulation and analysis
2. Scikit-Learn - Used for implementing and evaluating machine learning algorithms

Steps Taken

Step 1 - Future Men's Matches DataFrame Creation

In this step, I took the confirmed future men's matches schedule from ESPNcricinfo and used the Pandas library to convert it into a dataframe followed by converting it into a CSV for using for future predictions.

Step 2 - Loading the Datasets

I loaded the historical odi, tests and t20 dataset along with future men's matches dataset which was created in the previous step.

Step 3 - Cleaning Historical Data

In this step, I extracted the numerical values out of the margin columns in each of the historical odi, test and t20 datasets.

Step 4 - Calculating win ratios for each match type

Here, I added win ratios column to each historical dataset of odi, test and t20, by first doing the calculations using a function to clean and calculate the win ratio margin.

Step 5 - Preparing historical data with win ratios and margins

Here, I prepared each of the historical datasets by adding win ratios and margins to the dataset.

Step 6 - Training separate regression models for each match type

This step involves creating and evaluating a logistic regression model. Then we train the model.

Step 7 - Predicting Future Series Outcome

Here, the outcomes of future series matches is predicted for each match type based upon the features used in the previous steps.