See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/375884996

Predicting BPL Match Winners: An Empirical Study Using Machine Learning

Approach

Conference Paper · July 2023

DOI: 10.1109/ICCCNT56998.2023.10306698

CITATIONS READS

0 35

2 authors:

Bornita Adhikari Md. Sazzadur Ahamed

Daffodil International University Daffodil International University
1 PUBLICATION 0 CITATIONS 24 PUBLICATIONS 16 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Bornita Adhikari on 07 December 2023.

The user has requested enhancement of the downloaded file.

Predicting BPL Match Winners: An Empirical Study
Using Machine Learning Approach
Bornita Adhikari Md. Sazzadur Ahamed
Department. of Computer Science and Department. of Computer Science and
Engineering Engineering
Daffodil International University Daffodil International University
Dhaka,Bangladesh Dhaka,Bangladesh
[email protected] [email protected]

Abstract— With the evolution of computer science, every Every team want to give their best and for this purpose ML
company is implementing the newest technologies to survive in prediction can play significant role by handling any
market with better decision-making capabilities, better uncertainty and can make an impact in predicting the winner
communication and customer satisfaction. The only means of of matches using present data in several ways. ML models
fulfilling all these criteria’s is to perform data analysis that is can analyze the past data and predict how weather condition
more accurate and pure. In cricket, where no one can guess and a particular pitch or ground may affect the match’s
which team will win until the last ball of the last over, machine outcome. By analyzing the factors like team performance,
learning can help by predicting the results of the games. Match team ranking, head-to-head records and the recent form of
outcome prediction models have a lot of financial incentive
players, probability of winning the game at particular venue,
because cricket is a multi-billion-dollar industry. The goal of
score impact on field/bat first after winning the toss machine
this study is to identify the most accurate machine learning
model that can accurately predict the winner given the data learning models can predict which team is likely to win.
from the Bangladesh Premier League. For this analysis five ML In our analysis, the performance of each model and future
models XGBoost, Gradient Boosting, KNN, Decision Tree, directions are discussed with the goal of predicting the
Random Forest has been tested for the purpose of model outcome of BPL matches. With the growth of T20 leagues
building despite that our proposed model is XGBoost. To get and technological advancements the ask for cricket winner
access to BPL dataset web scrapping has been done, the dataset
prediction models is anticipated to expand in the upcoming
contains 15 columns and 3239 values and 8 team was available
years, as more fans and teams recognize the value of data-
in each season from 2018 to 2023. We use cutting-edge machine
learning techniques based on the use of numerous models,
driven insights and the potential competitive advantages they
feature selection, and data separation techniques. Finally, by can provide. Our experimental conclusion can help to
structuring every line of action, the forecast accuracy is optimize team strategies and increase their chances of
attained. winning. The prediction was convey using five machine
learning classifiers XGBoost, Gradient Boosting(GB), k-
nearest neighbors(KNN), Decision Tree(DT), Random
Forest(RF) where all the models in our research has shown
Keywords— BPL, Cricket, Prediction, XGBoost, outstanding accuracy. According to investigational result,
Visualization, Classification XGBoost indicates better prediction of 93%.We described the
benefits of cricket outcome prediction modeling with brief
introduction of this game. In section 2, some of the related
I. INTRODUCTION works about cricket outcome prediction is shown. The section
3, represent the workflow of this analysis, Section 4 comes
Cricket is a well-liked sport that is played and enjoyed by with the result formulation and conclusion section 5 of this
millions of people around the globe. It is especially popular in work provides a description.
countries such as India, Pakistan, Australia, England, South
Africa, Sri Lanka, Bangladesh, and the West Indies, where it
is considered a national pastime. The reason for the popularity
of cricket is it offers excitement and drama. The sport is II. LITERATURE REVIEW
known for its high-scoring matches, close finishes, and the
individual brilliance of its players. The fast-paced and Cricket has gained a lot of attention as it has progressed
dynamic nature of Twenty20 cricket has made it particularly among sports commentators. Cricket has been the subject of
popular in recent years, as it offers a shorter and more action- an increasing amount of research, but because this dataset is
packed version of the game. Nowadays Cricket is being brand-new and private, it has not yet been utilized in any
benefited from the growth of technology and the media, which research articles. Vistro et al. [1] conducted a study which
has made it more accessible to fans. Live streaming of
aimed to foresee the triumphant team in cricket matches
matches, social media, and mobile apps have made it easier
utilizing machine learning and data analytics methods. The
for fans to stay up-to-date with the latest news and scores.
study incorporated a range of features related to team and
A professional Twenty20 cricket league is called the player performance, venue, and other match-specific
Bangladesh Premier League (BPL) that was launched in 2012 variables to train their models. The study's findings showed
and it operates on franchise-based business model. BPL has that the suggested method able to foresee the winner of
become a t-twenty blast investing big amount of money. There cricket matches with an accuracy of more than 70%. The
are five individual winning team in BPL history, with the study accentuated the potential of machine learning and data
Dhaka Dynamites winning the most titles (3). Teams select analytics in predicting the winner of cricket matches. In the
their players based on draft system that says players should be study of Awan et al. [2] The team scores were predicted using
selected based on their performances in the previous season.

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

a machine learning linear regression model, and the accuracy Boosting, outperformed individual models, yielding match
was assessed using the big data framework Spark ML., outcome predictions that were over 60% accurate. In
RMSE, MSE, MAE on ODI dataset and got the outstanding Wickramasinghe's study [13], machine learning techniques
accuracy 95%. To predict the results of IPL matches, Pallavi were used to categorize all-rounders in One Day International
Tekade er al. [3] used several ML models, including Decision (ODI) cricket. The authors chose the fielding, bowling, and
Tree Regression, Random Forest Regression, Naive Bayes, batting statistics-related variables to train the algorithms. The
and Logistic Regression. They briefly described the key suggested method outperformed conventional statistical
factors, such as pitch conditions, temperature, humidity, and methods in categorizing all-rounders with high accuracy. The
precipitation, that directly affect the match scores. To study helps estimate cricket player performance using
illustrate their methods, they used a number of diagrams, and machine learning techniques. Overall, the study shows how
they recorded a 90% accuracy rate. With 70% accuracy, machine learning may be used to categorize and forecast
Vignesh Vapors Sankaranarayanan et al. [4] created a model cricket player performance. To find the ideal cricket match
to forecast the results of 125 ODI cricket matches. They also winning strategy, Srivastava et al. [17] suggested a hybrid
discuss how well the model performed. The difference machine learning-clustering-association rule architecture. In
between projected and actual total home runs, the overall order to train their models, the authors took advantage of a
score error distribution, and an examination of betting market variety of team composition, batting and bowling
sensitivity are all being shown using various data mining performance, and pitch characteristics factors. The best
approaches. Mittal and others [5] Kumar conducted a players and tactics for winning a match could be found using
thorough analysis and comparison of some of the most widely the suggested framework. The study adds to the body of
used machine learning algorithms, Naive Bayes, Logistic knowledge regarding the application of machine learning to
Regression Neural Networks, and Random Forest, among cricket strategy. In order to forecast player performance and
others, which virtually exactly predict the outcome of a team recommendations in the sport of cricket, Biswas et al.
match. Based on player performance and previous form, as [16] conducted a poll. The authors analyzed a number of
well as how many runs each batsman is expected to score and research that employed machine learning strategies to
how many wickets each bowler is expected to take, Kalpdrum forecast cricket player performance, team makeup, and game
Passi and Niravkumar Pandey [7] attempt to forecast the outcomes. The poll brought attention to how machine
outcome of the game. This document provides classification learning could enhance cricket team selection, training, and
reports of algorithms and a detailed computation of batting strategic decision-making.
and bowling attributes. In their investigation, the Random Using machine learning techniques, Ahmed et al.'s [14]
Forest classifier produced the highest accurate results. analysis of the Pakistan cricket team's performance. To
Kamble, R. R. [8] developed a system combining 2 models forecast game results and pinpoint variables that affect a
which can predict the score a team and prediction of the win team's success or failure, the authors examined several
percentage of both teams. This is an effective prediction performance indicators and player information. According to
system for cricket developed with 5 years of ODI data based the study, the performance of the Pakistani team can be
on Linear Regression, Naïve Bayes algorithm in their influenced by a number of variables, including player form,
analysis, Passi, Kalpdrum, and Niravkumar Pandey [9] used pitch conditions, and team composition. The study's
naive bayes, random forest, multiclass SVM, and decision conclusions show the potential of machine learning for
tree to create prediction models that could forecast how many cricket analytics and have practical ramifications for raising
runs a player would score and how many wickets a bowler the performance of the Pakistan side. This study adds to the
would take by analyzing attributes from ODI cricket matches. body of knowledge on cricket performance analysis and
The application of machine learning to forecast the results of strategic decision-making using machine learning. n their
English county twenty over cricket matches is explored in the 2017 study, Mustafa et al. sought to determine whether
study by Kampakis and Thomas (2015).The goal of this work machine learning techniques could accurately forecast cricket
by Lamsal and Choudhary [11] was to use machine learning match results based on user sentiment on social media. To
to predict the results of cricket matches in the Indian Premier find the most effective method for prediction, the authors
League.They put out a revolutionary strategy that blended compared various algorithms, including Random Forests
various machine learning algorithms and included both (RF), K-Nearest Neighbors (KNN), and Support Vector
simple and intricate features concerning team and player Machines (SVM). With a 67% accuracy rate, the data
performance, weather patterns, and other match-specific demonstrated that the RF algorithm performed better than the
aspects. The suggested model outperformed the benchmark others. The study emphasizes the potential of using online
model and had a prediction accuracy for matches of 64.1%. social network crowdsourcing to forecast sporting events.
The authors came to the conclusion that their strategy was an The investigation made by Mustafa et al. [6] (2017) aimed to
effective way to forecast cricket match results in the IPL and investigate the effectiveness of machine learning methods in
that it could also be used for other sports. Overall, this study predicting cricket match outcomes employing social network
adds to the body of knowledge on picking winners and losers user evaluations. The authors compared different algorithms
in cricket matches. such as Random Forests (RF), K-Nearest Neighbors (KNN),
Using machine learning methods, Srikantaiah, K. C. et al. and Support Vector Machines (SVM) to determine the best
[12] sought to forecast the results of Indian Premier League approach for prediction. The findings revealed that the RF
(IPL) matches. Authors use a variety of features to train their algorithm outperformed the others with an accuracy rate of
models, including batting and bowling statistics, location, 67%. The study highlights the potential of using crowd
and team performance. The findings demonstrated that opinions on social networks for predicting sports outcomes.
ensemble approaches, such as Random Forest and Gradient
 III. METHODOLOGY A. Data Description
To conduct the analysis, we sourced data from
cricbuzz.com that pertains to matches played in the BPL
Twenty20 cricket league from 2018 to 2023. Each season
comprises 46 matches and the data includes various attributes
such as the city, date, teams, toss winner, toss decision, winner
team, win-by-runs, result, win-by-wickets, and venue. To
ensure high accuracy and better outcomes when applying
prediction models, a significant amount of data was extracted,
resulting in a dataset that contains 15 columns and 3239
entries.
B. Data Pre-Processing
Preparing data for analysis involves several crucial steps,
collectively known as data pre-processing. These include
cleaning, transforming, and organizing raw data to make it
suitable for analysis. This step is crucial because the accuracy
and effectiveness of any analysis or modeling depend heavily
on the quality of the data. The various steps involved in pre-
processing are discussed in detail below.
 Removing Unwanted Columns
We eliminated null values and unnecessary columns like
ID, Season, and Man of the Match from this analysis in order
to focus on the columns that our forecast is founded on and
rely upon.
 Tackle Data Volume
To tackle the low volume of data causing lower
performance of algorithms, the data had to be volume up by
duplicating data rows.
 Label Encoding
Label encoding is the process of giving several levels or
categories of a variable or attribute a numerical value or code.
Fig. 1. Workflow of Methodology The pre-processing step of categorical data representation is
crucial for statistical modeling and data analysis. Data for ML
The workflow of this research are briefly elaborated in models must only be in numerical format. Therefore, label
Figure 1. The prime decision of this analysis was select encoding has been used to transform every entity into a
Bangladesh Premier League as our study work. First and numerical representation. Before label encoding, Table 1
foremost, data extraction was done. Before classification a displays all the properties and values of our dataset. The idea
small group of data preprocessing was needed. After testing of label encoding is displayed in Table 2.
several models, the proposed model was found, that got
highest accuracy.

TABLE I. BEFORE LABEL ENCODING

City date team1 Team2 toss- Toss- result winner Win-by- Win- value
runs by-
winner decision wickets
Chattogram 20/12/2019 Chattogram Comilla Comilla Field Normal 0 Zahur Ahmed
Challengers Warriors Warriors Chattogram Chowdhury
16
Challengers Stadium

Dhaka 11/1/2020 Dhaka Khulna Khulna Field Normal Khulna 0 8 Sher-e-Bangla

Platoon Tighers Tighers Tighers National Cricket
Stadium

Dhaka 11/1/2020 Dhaka Khulna Khulna Field Normal Dhaka 0 4 Sher-e-Bangla

Platoon Tighers Tighers Platoon National Cricket
Stadium

Dhaka 4/2/2018 Chattogram Dhaka Chattogram Bat Normal Dhaka 0 7 Sher-e-Bangla

Challengers Platoon Challengers Platoon National Cricket
Stadium

Dhaka 4/2/2023 Chattogram Comilla Chattogram Bat Normal Comilla 0 6 Sher-e-Bangla

Challengers Warriors Challengers Warriors National Cricket
Stadium
 TABLE II. AFTER LABEL ENCODING

City date team1 Team2 toss- Toss- result winner Win- Win- value
by-runs by-
winner decision wickets
0 41 0 2 1 1 1 0 16 0 4

2 8 2 5 4 1 1 6 0 8 1

2 79 2 5 4 1 1 2 0 4 2

2 75 0 3 0 0 1 2 0 7 2

2 76 0 2 0 0 1 1 0 6 2

C. Visualize dataset
 scatter plot matrix  visualize the correlation matrix Heatmap
In machine learning research, data visualization plays a To view the correlation matrix Heatmap, one needs to
crucial role in comprehending the connections between visualize a table that displays the correlation coefficients
variables, recognizing patterns, and gaining insights into between pairs of variables in a dataset. These coefficients are
model performance. One such visual representation is the statistical measures that indicate the degree of correlation
scatter plot matrix of features in xtrain, which displays the between two elements. Correlation coefficient values vary
correlations between pairs of features (or variables) in the from -1 to 1, where -1 denotes a fully negative correlation, 1
training dataset. To predict the winner of BPL matches, this denotes a fully positive correlation, and 0 denotes no
research has selected five columns - city, date, team1, team2, association. The plot that results from this visualization is a
and toss winner - to plot on the two axes, X and Y in a grid of matrix that is color-coded, with red hues representing positive
scatter plots. Each plot within the grid demonstrates the relationships and blue tones representing negative
relationship between two features, while the diagonal of the correlations. This display is shown in Figure3.
grid showcases the distribution of each feature. The
correlations observed in the dataset is shown in Figure 2

Fig. 2. Scatter plot matrix Fig. 3. Correlation matrix Heatmap

D. Classification models training probability or class label. No information has been omitted in
To prepare the models and preprocess the data, Google the paraphrased text.
Colaboratory's cloud GPU, along with essential libraries such E. Making predictions
as pandas, numpy, matplotlib, and seaborn, have been
employed. The data set includes a substantial number of After testing several model / machine learning algorithms,
entities, and the relevant variables that are likely to aid in it is seen that the models are getting trained conveniently. But
predicting the target variable (winner) have been selected the testing capability is not performed yet. To find out how the
based on the data description. Following the selection of algorithms are performing on each test portions of the dataset,
algorithms, the input data with corresponding target values we had to evaluate the model's performance through a step of
was fed into the model training. To increase the volume of prediction performed from the trained models separately and
data, the data was duplicated by a factor of two, and during the testing results indicate XGBoost model to be the highest
model training, the data was randomly divided into testing and performing one.
training sections. The training component used 80% of the
data, and the testing portion used the remaining 20%. Brief
descriptions of the tested models are provided below.
 KNN IV. RESULT AND DISCUSSION
The K-Nearest Neighbors (KNN) is a straightforward The accuracy table is helpful for assessing the
machine learning technique that is utilized for both regression effectiveness of models and pinpointing potential
and classification assignments. The algorithm functions by improvement areas. By emphasizing particular sorts of model
finding the K closest data points to a provided input data point flaws, the presentation of the accuracy table can help reveal
within the training set and then utilizing their values or labels areas that need work. The proportion of correctly classified
to predict an outcome. observations in the dataset after testing several models is
shown in Table 3 and Figure 4 is the proper visualization of
 Decision Tree the result.
The Decision Tree algorithm functions by dividing data TABLE III. MODEL ACCURACY
into smaller, similar subsets based on its defining features. To
make predictions, the input data is moved through the tree Models Accuracy
from the top node to the leaf node. The algorithm determines
XGBoost 93%
which branch to follow at each internal node by evaluating the
input feature's value. This process continues until the leaf node Gradient Boosting 92%
is reached, which provides a forecast for the input data. No
information has been left out in the paraphrased text. Random Forest 86%

 Random Forest Decision Tree 86%

One sort of ensemble learning that uses the Random Forest KNN 79%
method utilizes numerous decision trees to predict outcomes.
In order to make a prediction using this algorithm, the input
data is processed through each decision tree from the root
node to a leaf node. Once the data has reached the leaf node
of each tree, the algorithm produces a prediction based on
either the majority class or the average prediction value of all
the decision trees within the Random Forest.
 Gradient Boosting
Gradient Boosting is a technique that involves combining
several weak learners to form a powerful one. To make
predictions, the input data is first fed into the weak learner,
and the errors in the initial prediction are determined. To
improve the accuracy of the prediction, a new weak learner is
then trained to correct the previous learner. The final
prediction is obtained by adding up the predictions from all
the weak learners.
 XGBoost
XGBoost is a well-known machine learning model that is
widely used for regression, classification, and ranking tasks.
It is an ensemble model that consists of multiple decision trees
and leverages the errors of previous trees to enhance its
predictions. During the prediction process, the input data is
processed through various decision trees, and the scores
generated by each tree are merged to produce a final
prediction. Finally, a non-linear function like the sigmoid or
softmax function is applied to convert the output into a Fig. 4. Accuracy compares
 the training data. Predictions were then made on the test data
using the predict method. The evaluation of our model's
performance was done using accuracy as the primary metric.
In machine learning, performance metrics are
used to assess the effectiveness of different models. Accuracy
is a useful metric that measures the proportion of correctly
predicted instances in the dataset. Our XGBoost Classifier
achieved an accuracy of 94.25%, indicating a satisfactory
performance of our model.

C. Classification report of XGBoost

Table 4 presents a report
that summarizes the performance of various classes or
categories in the XGBoost dataset. The report includes key
metrics such as Precision, which represents the proportion of
predicted positives that are true positives, and Recall, which
refers to the proportion of actual positives that are correctly
identified. The F1 score is a combined measure of precision
and recall, while Support indicates the number of observations
in each class.

TABLE IV. AFTER LABEL ENCODING

Class Precision Recall F1-Score Support

Fig. 5. Confusion Matrix

Fortune 0.90 1.00 0.95 18

Barishal

A. Confusion Matrix:
The counts of true positives, true negatives, false positives, Khulna 0.91 1.00 0.95 20
and false negatives for each class in the dataset are shown in Tighers
the confusion matrix. In case the model is producing a lot of
false positives, we can attempt to enhance its performance by
modifying the decision threshold or experimenting with Comilla 1.00 1.00 1.00 7
different features. Figure 5 illustrates the confusion matrix of Warriors
the models that were tested.

Chattogram 1.00 1.00 1.00 2

B. Model Implementation –XGBoost Classifiers
Challengers

Extreme Gradient Boosting

(XGBoost) is a highly effective machine learning algorithm Dhaka 1.00 1.00 1.00 13
known for its remarkable predictive accuracy. It is specially Platoon
designed to be computationally efficient and capable of
handling large datasets. The algorithm incorporates L1 and L2
regularization, as well as dropout regularization, to prevent
overfitting and improve the model's generalization Sylhet 1.00 0.78 0.88 9
Thunder
performance. Additionally, XGBoost generates feature
importance scores, which help in identifying the most
significant features in the data. This feature can be utilized for
feature selection, feature engineering, and to gain insights into Rajshahi 0.83 0.71 0.77 14
various machine learning tasks such as regression, Royals
classification, and ranking.
The flexibility and customization of the
XGBoost algorithm is due to its support for various loss Rangpur 1.00 1.00 1.00 3
functions and evaluation metrics. To implement this Riders
algorithm, we utilized the BPL dataset from scikit-learn and
split it into training and testing sets. Our XGBoost model was
defined using the XGBClassifier class and was trained using
 V. CONCLUSION AND FUTURE WORK learning." Computer Science & Information Technology (CS & IT)
(2017).
Predicting the match winner of BPL data from season [10] 10. Kampakis, Stylianos, and William Thomas. "Using machine
2018 to 2023 was the primary goal of this study. We compared learning to predict the outcome of english county twenty over cricket
the performance of 5 different machine learning models to see matches." arXiv preprint arXiv:1511.05837 (2015).
which one best fits the dataset and The fields of data science [11] 11. Lamsal, Rabindra, and Ayesha Choudhary. "Predicting outcome
that have merged include visualizations, pre-processing, of Indian premier league (IPL) matches using machine learning." arXiv
implementing machine learning models, feature selection for preprint arXiv:1809.09813 (2018).
winner prediction and analysis. By Level encoding into [12] 12. Srikantaiah, K. C., Aryan Khetan, Baibhav Kumar, Divy Tolani,
and Harshal Patel. "Prediction of IPL match outcome using machine
numerical format, removing the missing values from the learning techniques." In 3rd International Conference on Integrated
dataset the preprocessing has been done By combining the Intelligent Computing Communication & Security (ICIIC 2021), pp.
target variable with the properties of the data, the best features 399-406. Atlantis Press, 2021.
were chosen. We have found outstanding results by [13] 13. Wickramasinghe, Indika. "Classification of all-rounders in the
performing. models for machine learning with specific game of ODI cricket: Machine learning approach." Athens Journal of
attributes. Our goal was to predict the winner with highest Sports 7, no. 1 (2020): 21-34.
accuracy, we achieved the highest accuracy on XGBoost and [14] 14. Dhonge, Nikhil, Shraddha Dhole, Nikita Wavre, Mandar
Pardakhe, and Amit Nagarale. "IPL cricket score and winning
predicted the winner with 93% accuracy. In this process of prediction using machine learning techniques." Int Res J
predicting the winner KNN gave us the lowest accuracy which Modernization Eng Technol Sci 3 (2021): 1723-1730.
is 79%. [15] 15. Ahmed, Waqar, Mahwish Amjad, K. Junejo, Tariq Mahmood,
and A. Khan. "Is the performance of a cricket team really
Because of different ways of data preprocessing, hyper unpredictable? a case study on pakistan team using machine learning."
parameter tuning, Sensitivity to initial conditions a single Indian Journal of Science and Technology 13, no. 34 (2020): 3586-
algorithm can produce different result. Therefore, we are 3599.
currently attempting to decrease the prediction inaccuracy. [16] 16. Biswas, Milon, Tajim Md Niamat Ullah Akhund, Md Kawsher
Furthermore, we intend to predict fall of wickets, finding Mahbub, Sikder Md Saiful Islam, Sadia Sorna, and M. Shamim Kaiser.
about the reasons behind most preferred decision of BPL after "A survey on predicting player’s performance and team
recommendation in game of cricket using machine learning." In
toss Field or Ball. Eventually, we aim to host a website where Information and Communication Technology for Competitive
all this analysis will be available for audience. Strategies (ICTCS 2020) ICT: Applications and Social Interfaces, pp.
223-230. Springer Singapore, 2022.
[17] 17. Srivastava, Praveen Ranjan, Prajwal Eachempati, Ajay Kumar,
Ashish Kumar Jha, and Lalitha Dhamotharan. "Best strategy to win a
match: an analytical approach using hybrid machine learning-
clustering-association rule framework." Annals of Operations
Research (2022): 1-43.
REFERENCES

[1] Vistro, Daniel Mago, Faizan Rasheed, and Leo Gertrude David. "The
cricket winner prediction with application of machine learning and data
analytics." International Journal of Scientific & Technology Research
8, no. 09 (2019).
[2] 2. Awan, Mazhar Javed, Syed Arbaz Haider Gilani, Hamza
Ramzan, Haitham Nobanee, Awais Yasin, Azlan Mohd Zain, and
Rabia Javed. "Cricket match analytics using the big data approach."
Electronics 10, no. 19 (2021): 2350.
[3] 3. Tekade, Pallavi, Kunal Markad, Aniket Amage, and Bhagwat
Natekar. "Cricket match outcome prediction using machine learning."
International journal 5, no. 7 (2020).
[4] 4. Sankaranarayanan, Vignesh Veppur, Junaed Sattar, and Laks VS
Lakshmanan. "Auto-play: A data mining approach to ODI cricket
simulation and prediction." In Proceedings of the 2014 SIAM
international conference on data mining, pp. 1064-1072. Society for
Industrial and Applied Mathematics, 2014.
[5] 5. Mittal, Harsh, Deepak Rikhari, Jitendra Kumar, and Ashutosh
Kumar Singh. "A study on machine learning approaches for player
performance and match results prediction." arXiv preprint
arXiv:2108.10125 (2021).
[6] 6. Mustafa, Raza Ul, M. Saqib Nawaz, M. Ikram Ullah Lali,
Tehseen Zia, and Waqar Mehmood. "Predicting the cricket match
outcome using crowd opinions on social networks: A comparative
study of machine learning methods." Malaysian Journal of Computer
Science 30, no. 1 (2017): 63-76.
[7] 7. Passi, Kalpdrum, and Niravkumar Pandey. "Increased prediction
accuracy in the game of cricket using machine learning." arXiv preprint
arXiv:1804.04226 (2018).
[8] 8. Kamble, R. R. "Cricket score prediction using machine learning."
Turkish Journal of Computer and Mathematics Education
(TURCOMAT) 12, no. 1S (2021): 23-28.
[9] 9. Passi, Kalpdrum, and Niravkumar Pandey. "Predicting players’
performance in one day international cricket matches using machine
View publication stats