IPL SCORE PREDICTION

USING MACHINE LEARNING

Dr K Ramesh babu*, Rachakonda Pujitha **, Padabakkala Sowmya *** .
Turaka Bhakthavatsalam ****, Vintha Preethi *****
CSE Department, GVR&SCET, JNTUK, Andhra Pradesh, INDIA.

ABSTRACT
cricket is the most popular game. The one of the most popular sports in the world,
Indian Premier League (IPL) is one of the redefining our perceptions and experiences
several series that are contested in the of the game. The Indian Premier League
nation. A model with two techniques has (IPL) is the centre piece of this cricket
been proposed. The first is a scoring revolution; it is a spectacular display of
prediction, and the second is a prediction of skill, cunning, and pure enjoyment.
the team winning. Linear regression, Machine Learning (ML) presents itself as a
logistic regression, decision trees, random game-changer in the ongoing quest to
forests, gradient boosting regressors, extra forecast match outcomes as the cricketing
tree regressors, and XGB regressors are globe changes. The use of machine learning
employed in these for score prediction. This techniques to sports analytics has created
study gathers and analyses IPL data new opportunities in recent years for
spanning multiple years, including player, analysing and predicting match results.
match, team, and ball-to-ball information, Cricket's unpredictable nature, particularly
to generate several conclusions that help in the T20 format of the IPL, offers machine
improve a player's performance. To learning algorithms a difficult but
forecast the winner, the model employed a fascinating platform to demonstrate their
supervised machine learning technique. ability to predict. Combining sports and
For high accuracy, Extra tree regressor technology together promises to improve
used for good accuracy with 90 % . spectator experiences while also providing
teams, coaches, and analysts with insightful
Keywords: Linear Regression, logistic
information on the dynamics of games. This
regression, decision tree, random forest,
paper explores the fascinating field of
gradient boosting regressor, extra tree
machine learning-based IPL score
regressor, XGB regressor IPL Winning
prediction. To identify the patterns and
Prediction, IPL Score Prediction, ball-to-
trends that affect match outcomes, we use a
ball Statistics.
variety of characteristics, including pitch
1.INTRODUCTION conditions, player statistics, historical data,
and other variables. We want to explain the
The introduction of technology has brought
intricacies of the game and offer a view into
about a revolutionary alteration in cricket,
the future, one in which projecting IPL
scores will no longer be only a speculative

174
© 2024, IRJEdT Volume: 06 Issue: 05 | May -2024
endeavour, thanks to machine learning. in order to predict the result of a match
2.LITERATURE SURVEY using player performance data and the
history of IPL cricket .[9] The accuracy of
[1]Using machine learning techniques like
Linear Regression in Score Prediction
Decision Tree, SVM, Decision Tree,
Analysis is higher than that of Ridge and
logistic regression, random forest classifier,
Lasso Regression .[10] In this study, we
and k-nearest, the authors of discovered
present the feedback analysis for tweets
and noted some things. The random forest
following IPL-2020 matches and examine
classifier beats every algorithm in this
the team's level of fame during the
experiment by accurately predicting the
competition.
outcome with the highest accuracy of
88.10%.[2]This work has examined and 3. SUMMARY OF LITERATURE
analysed IPL score prediction in SURVEY
Understanding the IPL data set from the
Studying all of the aforementioned
previous ten years is the goal of this publications revealed gaps in knowledge,
endeavour. Understanding the operation limits in current methods, and lower
and use of the four distinct machine accuracy when compared to one another.
learning algorithms is beneficial.[3] work Their obtained results show a lack of
utilizing machine learning algorithms in accuracy. Furthermore, some algorithms to
Each player's point total was utilized to estimate IPL score prediction encounter
difficulties due to regional diversity in some
determine each team's relative strength.
publications. The dataset may have an
Using the IPL dataset created for this impact on the accuracy. The following
purpose, several classification-based regressors were employed in this paper:
machine learning algorithms were trained. gradient boosting, extra tree, XGB,
In [4] The research focused on predicting decision tree, logistic regression, and linear
the winner for an IPL match using machine regression.
learning and utilizing the available
4.PROPOSED ARCHITECTURE
historical data of IPL from season 2008-
2019.In [5] This paper will give the
important information regarding IPL score
prediction and winning prediction system,
that which parameters are required also the
classifiers and algorithms. In [6] This will
make things easier so that anyone checks
the match prediction just by using their
mobile or PC. The proposed LR algorithm
shows better results as compared to the
other previous ML algorithms. [7]. When
the actual scores and the predictions were Fig 4.1 IPL score prediction Architecture
compared in , the findings showed a strong Data collection
association between the two. [8] the In order to capture the several aspects
average impact factor of the team based on influencing match outcomes, a
featuring players is taken into consideration comprehensive strategy is used in the data

175
© 2024, IRJEdT Volume: 06 Issue: 05 | May -2024
collection process for IPL score prediction.
First, a compilation of past match data from
several IPL seasons is made, which
includes information on team lineups,
batting orders, individual player
performances, pitch and weather
conditions, venue statistics, and match
outcomes. From the fig 4.1 IPL score for binary classification problems, where
shown the predictive models are trained the outcome variable y is categorical and
using this historical dataset, which helps Fig 4.2 logistic regression
them find patterns and connections between
various factors and match results. Up-to- has only two possible outcomes (0 and 1, or
date information on player injuries, team "yes" and "no"). It models the probability
tactics, and other dynamic aspects that a given input belongs to a certain
category. For example, in medical
preceding each match is also incorporated
diagnosis, logistic regression can be used to
using real-time data feeds. Sources of predict whether a patient has a particular
structured data include official IPL sources, disease or not based on various features
statistical databases, and cricket analytics such as age, gender, blood pressure, etc.
platforms; news outlets and social media ii). Decision tree Fig 4.3 shows below
offer additional perspectives on public explain A decision tree is a machine
learning algorithm used for both
opinion and team dynamics. The predictive
classification and regression tasks. It
models can be trained to provide educated models decisions and their possible
predictions about team scores in upcoming consequences by creating a tree-like
IPL matches by combining and structure of decisions, making it an intuitive
preprocessing this wide range of data and visually interpretable method. Decision
sources. This would enable data-driven trees split the data into branches to make
predictions, using the structure of a tree
decision-making for stakeholders in the
consisting of nodes and leaves. A Decision
cricket ecosystem. Tree is a versatile and intuitive machine-
ALGORITHMS USED learning algorithm used for both
In machine learning mainly there are two classification and regression tasks. It's a
types of algorithm models they are: tree-like structure where an internal node
Classification Algorithms, Regression represents a feature (or attribute), the
Algorithms. branch represents a decision rule, and each
Classification Algorithms leaf node represents the outcome (or class
Classification algorithms are used to predict label). The below diagram explains the
categorical labels or classes for new data general structure of a decision tree:
points based on past observations. The goal
is to learn a mapping from input features to
predefined categories.
i). Logistic Regression Fig 4.2 shows
below explain Logistic regression is a
statistical method and a type of predictive
analysis used in machine learning tasks.
Logistic regression statistical method used

176
© 2024, IRJEdT Volume: 06 Issue: 05 | May -2024
 predictions on data. Linear regression is
also a type of machine-learning algorithm
more specifically a supervised machine-
learning algorithm that learns from the
labelled datasets and maps the data points
to the most optimized linear functions.
which can be used for prediction on new
datasets

Fig: 4.3 Decision tree algorithms

iii). Random Forest Fig 4.4 shows below
explain A Random Forest is an ensemble
learning method used in machine learning
that operates by constructing a multitude of
decision trees at training time and
outputting the class that is the mode of the
classes (classification) or mean prediction
(regression) of the individual trees.
Random Forest is an ensemble learning Fig 4.5 linear regression
method used for both classification and
regression tasks. It operates by constructing v). Gradient boosting regressor Gradient
a multitude of decision trees during training boosting is a machine learning ensemble
and outputs the class that is the mode of the technique that combines the predictions
classes (classification) or the mean of multiple weak learners, typically
decision trees, sequentially. It aims to
improve overall predictive performance
by optimizing the model’s weights based
on the errors of previous iterations,
gradually reducing prediction errors and
enhancing the model’s accuracy.

Fig: 4.4 Random Forest algorithms

prediction (regression) of the individual

trees. Here is an explanation of the Random
Forest algorithm: The below diagram
explains the working of the Random Forest.

iv). Linear regression Machine

Learning is a branch of Artificial
intelligence that focuses on the
development of algorithms and statistical Fig 4.6 gradient boosting regressor
models that can learn from and make

177
© 2024, IRJEdT Volume: 06 Issue: 05 | May -2024
vi).Extra tree regressor an extra-trees

regressor. This class implements a meta

Fig 4.8 XGB regressor
estimator that fits several randomized
decision trees (a.k.a. extra-trees) on various 5. RESULTS
sub-samples of the dataset and uses
averaging to improve the predictive
accuracy and control over-fitting. Read
more in the User Guide. The number of
trees in the forest.

Fig 4.9 executed results comparison.

from the above it is compared that the

seven machine algorithms are used with
Fig 4.7 extra tree regressor the code execution results.
vii). XGB regressor Extreme Gradient
Boosting (XGB Boost) is an open-source
library that provides an efficient and
effective implementation of the gradient
boosting algorithm. Shortly after its
development and initial release, XGB
Boost became the go-to method and often
the key component in winning solutions for
a range of problems in machine learning
competitions.

Fig 4.10 accuracy comparison table

178
© 2024, IRJEdT Volume: 06 Issue: 05 | May -2024
From the above comparison table it and dependability of predictions. By
concluded that the extra tree regressor gives combining the advantages of several
the best accuracy of 90%. models, ensemble approaches can mitigate
the shortcomings of individual models and
6. CONCLUSION offer more thorough insights on match
results.
The goal of this is to use machine learning
techniques to forecast the scores of IPL 8. REFERENCE
matches. it gives a prediction model with [1] S. M, E. R. G, G. P, J. M, K. K and S.
encouraging outcomes by carefully sk. V, "A Novel Approach to Predicting
preprocessing the data, creating features, IPL Player Value and Score Using XGB
and choosing the model. The examination Boost and Ridge Regression," 2023 6th
showed that a number of important International Conference on Recent Trends
variables, including individual statistics, in Advance Computing (ICRTAC),
past data, venue, team performance, and the Chennai, India, 2023, pp. 343-348, Doi:
result of the toss, greatly affect match 10.1109/ICRTAC59277.2023.10480773
scores. further improve forecast accuracy,
[2] K. Suresh, B. Vikas, Kanishka and K.
future works could look into adding
Vikas, "Design and Analysis of a Chatbot
features like player injuries, match
with IPL First Inning Score
dynamics, and weather in real-time.
Prediction," 2021 International
Advanced machine learning methods like
Conference on Advancements in Electrical,
XGB regressor, gradient boosting
Electronics, Communication, Computing
regressor, extra tree regressor, logistic
and Automation (ICAECA), Coimbatore,
regression, decision tree, random forest,
India, 2021, pp. 1-4, Doi:
and linear regression are used. As the model
10.1109/ICAECA52838.2021.9675645.
shows how machine learning may be used
to forecast IPL match scores, from the all [3] E. Mundhe, I. Jain and S. Shah, "Live
algorithms here used extra tree regressor Cricket Score Prediction Web Application
which gave the best accuracy of 90%. using Machine Learning," 2021
International Conference on Smart
7. FUTURE SCOPE
Generation Computing, Communication
There are many opportunities for research and Networking (SMART GENCON),
and development in the broad and exciting Pune, India, 2021, pp. 1-6, Doi:
field of IPL score prediction using machine 10.1109/SMARTGENCON51891.2021.96
learning. One area of interest is improving 45855.
current models to include more variables
and data sources, like team dynamics, [4]T. Bhalerao, S. Vijayalakshmi and G. J,
player tiredness, and even sentiment "A Comparative Analysis On Machine
analysis from social media sites for fans. Learning Algorithm for Score Prediction
Through the integration of these
parameters, models are able to provide and Proposal of Enhanced Naïve
more accurate and forecasts, better Bayes," 2022 4th International Conference
reflecting the awareness of cricket matches. on Advances in Computing,
Additionally, there's a chance to create Communication Control and Networking
ensemble models that mix many methods (ICAC3N), Greater Noida, India, 2022, pp.
and algorithms to improve the robustness
179
© 2024, IRJEdT Volume: 06 Issue: 05 | May -2024
618-621, Doi: Learning," 2023 7th International
10.1109/ICAC3N56670.2022.10074063. Conference On Computing,
Communication, Control And Automation
[5] Y. Kumar, H. Sharma and R. Pal,
(ICCUBEA),Pune,India,2023,pp.1-6,doi:
"Popularity Measuring and Prediction
10.1109/ICCUBEA58933.2023.10392103.
Mining of IPL Team Using Machine
Learning," 2021 9th International [10] E. Lug Hofer et al., "Evolving time-
Conference on Reliability, Infocom series based prediction models for quality
Technologies and Optimization (Trends criteria in a multi-stage production
and Future Directions) (ICRITO), Noida, process," 2018 IEEE Conference on
India, 2021, pp. 1-5, Doi: Evolving and Adaptive Intelligent Systems
10.1109/ICRITO51393.2021.9596405. (EAIS), Rhodes, Greece, 2018, pp. 1-10,
Doi: 10.1109/EAIS.2018.8397186.
[6] T. Bhalerao, S. Vijayalakshmi and G. J,
"A Comparative Analysis On Machine
Learning Algorithm for Score Prediction
and Proposal of Enhanced Naïve
Bayes," 2022 4th International Conference
on Advances in Computing,
Communication Control and Networking
(ICAC3N), Greater Noida, India, 2022, pp.
618-621, Doi:
10.1109/ICAC3N56670.2022.10074063.

[7]K. Ishwarya and A. A. Nithya, "Relative

Analysis and Performance of Machine
Learning Approaches in Sports," 2021 5th
International Conference on Electronics,
Communication and Aerospace
Technology (ICECA), Coimbatore, India,
2021, pp. 1084-1089, Doi:
10.1109/ICECA52323.2021.9676147.

[8]S. G, A. Swaminathan, J. B. J, S. R and

L. Nelson, "IPL Data Analysis and
Visualization for Team Selection and Profit
Strategy," 2023 7th International
Conference on Computing Methodologies
and Communication (ICCMC), Erode,
India, 2023, pp. 592-598, Doi:
10.1109/ICCMC56507.2023.10083736.

[9] D. JaVale, N. Potnis, A. Koli, M.

Bharata and B. Tiple, "Impact of Toss on
the Result of IPL Matches Using Machine

180
© 2024, IRJEdT Volume: 06 Issue: 05 | May -2024