Heart Disease Prediction Using Machine Learning Techniques

Abstract

Machine learning and artificial intelligence have been found useful in various disciplines during the course of their development, especially in the enormous increasing data in recent years. It can be more reliable for making better and faster decisions for disease predictions. So, machine learning algorithms are increasingly finding their application to predict various diseases. Constructing a model can also help us visualize and analyze diseases to improve reporting consistency and accuracy. This article has investigated how to detect heart disease by applying various machine learning algorithms. The study in this article has shown a two-step process. The heart disease dataset is first prepared into a required format for running through machine learning algorithms. Medical records and other information about patients are gathered from the UCI repository. The heart disease dataset is then used to determine whether or not the patients have heart disease. Secondly, Many valuable results are shown in this article. The accuracy rate of the machine learning algorithms, such as Logistic Regression, Support vector machine, K-Nearest-Neighbors, Random Forest, and Gradient Boosting Classifier, are validated through the confusion matrix. Current findings suggest that the Logistic Regression algorithm gives a high accuracy rate of 95% compared to other algorithms. It also shows high accuracy for f 1-score, recall, and precision than the other four different algorithms. However, increasing the accuracy rates to approximately 97% to 100% of the machine learning algorithms is the future study and challenging part of this research.

Full text

American Journal of Computer Science and Technology
2022; 5(3): 146-154
http://www.sciencepublishinggroup.com/j/ajcst
doi: 10.11648/j.ajcst.20220503.11
ISSN: 2640-0111 (Print); ISSN: 2640-012X (Online)
Heart Disease Prediction Using Machine Learning
Techniques
Mohammed Khalid Hossen
Department of Computer Science and Engineering, Sylhet Agricultural University, Sylhet, Bangladesh
Email address:
To cite this article:
Mohammed Khalid Hossen. Heart Disease Prediction Using Machine Learning Techniques. American Journal of Computer Science and
Technology. Vol. 5, No. 3, 2022, pp. 146-154. doi: 10.11648/j.ajcst.20220503.11
Received: June 27, 2022; Accepted: July 13, 2022; Published: July 20, 2022
Abstract:
Machine learning and artificial intelligence have been found useful in various disciplines during the course of their
development, especially in the enormous increasing data in recent years. It can be more reliable for making better and faster
decisions for disease predictions. So, machine learning algorithms are increasingly finding their application to predict various
diseases. Constructing a model can also help us visualize and analyze diseases to improve reporting consistency and accuracy.
This article has investigated how to detect heart disease by applying various machine learning algorithms. The study in this article
has shown a two-step process. The heart disease dataset is first prepared into a required format for running through machine
learning algorithms. Medical records and other information about patients are gathered from the UCI repository. The heart disease
dataset is then used to determine whether or not the patients have heart disease. Secondly, Many valuable results are shown in this
article. The accuracy rate of the machine learning algorithms, such as Logistic Regression, Support vector machine, K-Nearest-
Neighbors, Random Forest, and Gradient Boosting Classifier, are validated through the confusion matrix. Current findings suggest
that the Logistic Regression algorithm gives a high accuracy rate of 95% compared to other algorithms. It also shows high
accuracy for f
1
-score, recall, and precision than the other four different algorithms. However, increasing the accuracy rates to
approximately 97% to 100% of the machine learning algorithms is the future study and challenging part of this research.
Keywords:
Machine Learning, Artificial Intelligence, Heart Disease, Linear Regression, Support Vector Machine,
K-Nearest-Neighbors, Random Forest, Decision Tree, Gradient Boosting
1. Introduction
Machine learning (ML) is a part of artificial intelligence
(AI) that allows a software application to improve its
prediction accuracy without being formally programmed. In
order to forecast new output values, machine learning
algorithms use historical data as input [1]. Machine learning
is a significant and diversified field, and its scope and
application are expanding daily. For this reason, machine
learning has become a crucial competitive differentiation in
many organizations. Machine learning includes supervised,
unsupervised, and ensemble learning classifiers that are used
to predict and find the accuracy of a dataset. ML algorithms
can build a model based on sample data called train data to
make a decision or prediction [1, 2].
The use of machine learning methods in the medical
industry is the subject of the current study, which mainly
focuses on mimicking some human activities or mental
processes and recognizing diseases from a variety of inputs [3].
The term “heart disease” refers to a group of conditions that
affect the heart. According to World Health Organization
reports, cardiovascular diseases are now the leading cause of
death worldwide, approximately 17.9 million [4, 5]. Many
types of research have been studied and performed with
various machine learning algorithms to diagnose heart diseases.
According to Ghumbre et al., machine learning and deep
learning algorithms are applied to predict heart diseases in the
UCI dataset [3]. The authors concluded that machine learning
algorithms performed better for this analysis. Machine learning
techniques for heart disease prediction are published by Rohit
Bharti et al., where the article concluded that different data
mining and neural system should be used to find the
seriousness of HD among patients [4]. Some analysis has been
led to think about the implementation of a predictive data
mining strategy on the same dataset [5]. Prediction of heart
disease using machine learning is studied by Jee S H et al. in

147 Mohammed Khalid Hossen: Heart Disease Prediction Using Machine Learning Techniques
which training and the testing dataset are performed by using a
neural network algorithm [6]. K-Nearest Neighbor algorithm is
reviewed to diagnose heart disease by Mai Shouman et al. [7].
Some efficient algorithms have been used to detect HD, which
shows results that each algorithm has its strength to register the
defined objectives [8]. The supervised network has been
applied for HD diagnosis, which is studied by Raihan M et al.
[9]. This research idea has been broadened and inspired us
worldwide by publishing many articles [10-15].
This article will construct an ML predictive model, which
will help analyze heart disease regarding the medical history.
Data is collected from the UCI repository with patients'
medical records and attributes. This dataset would be utilized
to predict whether the patients have heart disease or not. To
diagnose the HD dataset, this article considers 14 attributes of
a patient. It classifies whether the disease is present or not and
can help us diagnose diseases with fewer medical treatments [1,
5]. For this study, this article considers various attributes of
patients like age, sex, serum cholesterol, blood pressure, exang,
etc. Five different ML algorithms such as Logistic Regression
(LR), Support vector machine (SVM), K-Nearest-Neighbors
(KNN), Random Forest (RF), and Gradient Boosting Classifier
(GBC) are applied for the purpose of classification and
prediction of heart disease. Many beneficial results are
presented in this article. The attributes of the given dataset are
trained under these algorithms. Based on the characteristics of
the HD dataset, a comparative analysis of algorithms has been
studied regarding the accuracy rate. All the selected ML
algorithms are efficient by showing their accuracy, which is
greater than 80%. The most efficient algorithm is Logistic
Regression (LR), which gives us an accuracy rate of
approximately 95%. Finally, Logistic Regression (LR)
algorithm will be considered to predict and diagnose for heart
disease of a patient.
This article is rearranged sequentially. In section 2, the
methodology has been discussed. Various ML algorithms are
studied briefly in section 3. Results and analysis are shown in
section 4. In the result section, algorithms are compared
regarding the confusion matrix. Finally, a conclusion and
future scope have been drawn in section 5.
2. Methodology
In this section, the method and analysis are described,
which is performed in this research work. First of all, the
collection of data and selection of relevant attributes are the
initial steps in this study. After that, the relevant data is pre-
processed into the required format. The given data is then
separated into two categories: training and testing datasets.
The algorithms are then used, and the given data train the
model. The accuracy of this model is obtained by using the
testing data. The procedures of this study are loaded by using
several modules such as a collection of data, selection of
attributes, pre-processing of data, data balancing, and
prediction of disease.
2.1. Data Collection
In this article, the dataset is collected from the UCI
repository, which is considered in research analysis by the
many authors [4, 7]. So, the first step is organizing the
dataset from the UCI repository to predict the heart disease
and then dividing the dataset into two sections: training and
testing. In this article, 80% data has been considered as a
training dataset, and 20% dataset is used for testing purposes.
2.2. Dataset and Attributes
Attributes of a dataset are properties of a dataset, which
are important to analyze and make a prediction regarding our
concern. Various attributes of the patient, like gender, chest
pain, serum cholesterol, fasting blood pressure, exang, etc.,
are considered for predicting diseases. However, the
correlation matrix can be used for attribute selection to
construct a model.
Table 1. Attributes used are listed.
Sl. No. Attributes Description Values
1. Age Patients age in years Continuous
2. Sex Sex of subject (male-0, female-1) Male/Female
3. CP Chest pain type Four types
4. Trestbps Resting blood pressure Continuous
5. Chol Serum cholesterol in mg/dl Continuous
6. FBS Fasting blood pressure < or >120 mg/dl
7. Restecg Resting Electrocardiograph Five values
8. Thalach Maximum heart rate achieved Continuous
9. exang Exercise Induced Angina Yes/No
10. oldpeak ST Depression introduced by exer. Continuous
11. slope Slope of Peak Exercise ST segment up/flat/down
12. Ca Number of major vessels 0-3
13. thal Defect type Reversible/Fixed/Normal
14. Targets Heart disease 1 (disease), 0 (no disease)
2.3. Pre-processing of Data
We need to clean and remove the missing or noise values
from the dataset to obtain accurate and perfect results, known
as data cleaning. Using some standard techniques in python
3.8, we can fill missing and noise values, see [16]. Then we
need to transform our dataset by considering the dataset's
normalization, smoothing, generalization, and aggregation.

American Journal of Computer Science and Technology 2022; 5(3): 146-154 148
Integration is one of the crucial phases in data pre-
processing, and various issues are considered here to
integrate. Sometimes the dataset is more complex or difficult
to understand. In this case, the dataset needs to be reduced in
a required format, which is best to get a good result.
2.4. Balancing of Data
Balancing the dataset is necessary to improve the
performance of machine learning algorithms. A balanced
dataset has the same amount of input samples for each output
class (or target class). The imbalanced dataset can be
balanced by considering two methods, such as under
sampling and over sampling.
2.5. Prediction of Disease
In this article, five different machine learning algorithms
are implemented for classification. A comparative analysis of
the algorithms has been studied. Finally, this article considers
an ML algorithm that gives the highest accuracy rate for
heart disease prediction, see Figure 1.
Figure 1. Architecture of prediction models.
3. Machine Learning Algorithms
A data analysis technique called machine learning
automates the development of analytical models. In this
observation, five different algorithms are studied to obtain
the accuracy for finding the best one.
3.1. Logistic Regression Model
This ML model is often used for classification and
predictive analysis, also known as logit regression [16]. It is
also utilized to estimate the discrete values, like the binary
outcome, from a collection of independent variables. A
binary result means two possibilities will happen: either the
event happens (say 1), or it does not happen (say 0).
Here below are the working procedures of the Logistic
Regression model.
Figure 2. Logistic Regression model.
Where z is a function of x1, x2, w1, w2, and b. So, z is a
linear equation given to a sigmoid function to predict the
output. We calculate the loss to evaluate the performance of
this model. In this case, we use the cross-entropy loss

149 Mohammed Khalid Hossen: Heart Disease Prediction Using Machine Learning Techniques
function [17].
3.2. Support Vector Machine (SVM)
SVM is the most popular supervised machine learning
algorithm, which is used for classification as well as regression
[23]. Although, we primarily consider this algorithm for
classification problems in ML. The purpose of the SVM
algorithm is to construct the optimum decision boundary or
line that can divide n-dimensional space into classes so that we
can quickly put the new data point in the correct category. This
optimal decision boundary is known as hyperplane [23]. SVM
selects the extreme vectors that help to create the hyperplane.
The extreme vectors are known as support vectors, and the
algorithm with support vectors is called the support vector
machine. Here below is a figure of SVM, where the decision
boundaries or hyperplane classifies two different categories.
The training sample dataset is (x2, x1), where x1 is the x-axis
vector, and x2 is the target vector, see figure 3.
Figure 3. Support Vector Machine.
3.3. K-Nearest Neighbors (K-NN)
K-NN is the most straightforward classification algorithm
based on supervised learning techniques. However, the K-NN
algorithm can also be used for regression but is mostly used
for classification [18]. A new data point is classified by using
the K-NN algorithm depending on how similar the existing
data is stored. It indicates that the K-NN algorithm can
quickly classify new data when it appears in a suitable
category, see Figure 4.
Figure 4. K-Nearest Neighbors.
Here, the horizontal x-axis and vertical y-axis are
independent and dependent variables of a function,
respectively. Figure 4 is a simple example of the K-NN
classification algorithm. The test sample (Yellow Square with
what symbol) should be classified as either a green triangle
or a red star in this algorithm. When k=3 is considered in a
small dash circle, the yellow square would be a green triangle
because the majority number in this region is green triangles,
not red stars. Now, if we consider k=7, which is in a large
dash circle, then the yellow square would be red stars
because the number of red stars is four and the green
triangles are 3. So, It can conclude that the majority vote in a
specific region is important here, see Figure 5.
3.4. Random Forest
Random Forest (RF) is a popular supervised machine
learning algorithm used for both classifications and
regression. However, it is mainly used in classification
problems. RF algorithm is based on the concept of ensemble
learning. Ensemble learning is a general machine learning
procedure that can be used for multiple learning algorithms
to seek better predictive performance [2, 19]. So, the RF
technique creates several decision trees on the data samples,
obtains the prediction from each tress, and finally gets the
better solution by considering the majority voting. It is noted
that the ensemble method is better than a single decision tree
because it mitigates the over-fitting by averaging results. The
large number of decision trees in RF helps us to get the
accuracy and prevent over-fitting of the problems. The
following procedures are completed by RF algorithm, see
also figure 6:
Step 1: First, n numbers of the random sample are selected
from a given dataset.
Step 2: A decision tree will be constructed for every
individual.
Step 3: Each decision tree will predict an output.
Step 4: Final result has come through a majority voting or
averaging.
Figure 5. General procedure of Random Forest.
3.5. Gradient Boosting
Gradient Boosting (GB) is a machine learning technique

American Journal of Computer Science and Technology 2022; 5(3): 146-154 150
that is used in classification and regression problems like
others. It is a powerful algorithm in the field of machine
learning [21]. As is well known, the errors are classified into
two categories in machine learning algorithms: Bias error and
Variance error. GBC helps us to minimize bias error
sequentially in the model, see Figure 6. A diagram is
described as follows below,
Figure 6. Diagram Gradient Boosting (Source [22]).
As we can see that the ensemble consists of N trees; see
Figure 6. First of all, the feature matrix X and the labels y are
used to train Tree 1. To calculate the training set residual
error r1, the predictions labeled are used. Then, Tree 2 is
trained using feature matrix X and residual errors r1 of Tree 1
as labels. The residual error r2 is then calculated by using
predictive error, see Figure 6.
4. Result Analysis
4.1. Analysis of Heart Disease Dataset
Figure 7. Target class.
Before going to study the performance of considering
machine learning algorithms in this research, analysis of
the features of the heart disease dataset will be focused on
here. The total number of observations in the target
attributes is 1025, where not having heart disease 499
(denoted by 0) and having heart disease 526 (represented
by 1), see Figure 7. So, the percentage of not having heart
disease is 45.7%, and the percentage of having heart
disease is 54.3%, see Figure 8(a). It is shown that the rate
of heart disease is more than the rate of no heart disease.
In Figure 8(b), the sex feature of the HD dataset is
observed through the target feature. In sex attribute, the
female and male numbers are 312 and 713, respectively.
So, the male number is more than double of female
number. We can see in this figure 8(b) that the number of
heart diseases in males is higher than in females.
Similarly, no heart disease among males is higher than in
females. Figure 8(b) concludes that male is sufferer than
female; for more information, see figure 8(b).
Figure 8. (a) Percentage of no heart disease and heart disease, (b)
Comparison between sex and target feature.
Figure 9(a) shows a relationship between age and
cholesterol with the target feature. For an experiment, these
features from the dataset are considered randomly. The trend
of no heart disease is higher from 55 to 68 when the
cholesterol level is between 200 mg/dl and 300 mg/dl. For
validation, the KDE plot 9(b) is studied and shows similar
statistics.

151 Mohammed Khalid Hossen: Heart Disease Prediction Using Machine Learning Techniques
Figure 9. (a) Age v/s Cholesterol with the target feature, (b) Kernel density estimate (kde) plot of age v/s cholesterol.
The correlation of the features is drawn in figure 10. The
main purpose of the correlation plot is to define the positive
and negative correlation between the features. However, it
assumes that figure 10 is complex for getting the strong and
weak correlation. For this reason, this article added another
figure 11 to obtain these correlations efficiently. In figure 11,
we can see that three features like cp, thalach, and slope
positively correlate with target features.
Figure 10. Correlation matrix of the attributes.

American Journal of Computer Science and Technology 2022; 5(3): 146-154 152
Figure 11. Correlation with the target feature.
Two strong correlations by cp and slope with target feature
are studied statistically. As we can see in figure 12(a), there is
no heart disease when the cp level is more than 350;
however, heart disease is sustained more when the cp is
between 200 and 250. In addition, when the slope is in 300 <
slope-1 < 350, it shows that there is no disease, see figure
12(a). In contrast, for slope-2, there is a heart disease in 300
< slope-2 < 350.
Figure 12. (a) cp v/s target, (b) slope v/s target.

153 Mohammed Khalid Hossen: Heart Disease Prediction Using Machine Learning Techniques
4.2. Performance Analysis
In this article, various machine learning algorithms like
Logistic Regression (LR), Support vector machine (SVM), k-
Nearest-Neighbors (KNN), Random Forest Classifier (RF),
and Gradient Boosting Classifier (GBC) are studied broadly
to predict the heart disease. The accuracy rate of each
algorithm has been measured, and selects the algorithm with
the highest accuracy. The accuracy rate is a correct prediction
ratio to the total number of given datasets. It can be written
as,
Accuracy =

 
Where, TP: True Positive
TN: True Negative
FP: False Positive
FN: False Negative
After performing the machine learning algorithms for
training and testing the dataset, we can find the better
algorithm by considering the accuracy rate. The rate of
accuracy is calculated with the support of a confusion matrix.
As shown in Table 2, the Logistic Regression algorithm gives
us the best accuracy to compare with other ML algorithms.
Table 2. Accuracy comparison of algorithms.
Algorithms Accuracy
Logistic Regression (LR) 0.95
Support vector machine (SVM) 0.90
K-Nearest-Neighbors (KNN) 0.87
Random Forest Classifier (RF) 0.79
Gradient Boosting Classifier (GBC) 0.80
Figure 13. Accuracy comparison of machine learning algorithms by bar
diagram.
This has been studied more on the LR machine learning
algorithm through confusion matrix and f
1
-score. The
confusion matrix shows that the correct predicted value is
95%, see figure 14. f
1
-score is calculated by, which is shown
in figure 15,
f
1
= 2 *









Where, Precision, P =


, and Recall R =


.
Figure 14. Confusion matrix of LR algorithm.
Figure 15. f
1
-score, precision, and recall of LR algorithm.
5. Conclusion and Future Scope
The heart is a vital organ in the human body; however,
heart disease is a major concern in the world because this
disease is increasing day by day. So, we can handle this
disease if we have a model which can predict the initial
condition of heart disease. So, we need to create a machine
learning model that can be more accurate and help to
diagnose heart disease with less doubt and cost. It can be a
primary technique for knowing the condition of the heart. For
this reason, this article focuses on the heart disease prediction
based on the accuracy rate of the confusion matrix.
Following this idea, the statistics of the given algorithms are
used to estimate the accuracy rate of confusion matrix and
validated the statistics among the machine learning
algorithms. When five algorithms are compared, it is found
that Logistic Regression algorithm is selected regarding the
performance of high accuracy rate. The accuracy rate of
Logistic Regression model is 95%, which indicates that
machine learning algorithm will be considered as a pre-
defined tool to seek heart diseases in the near future. Other
statistics such as f
1
-score, recall, and precision rate have been
calculated for Logistic Regression as 95%, 95%, and 95%,
respectively. These estimated values suggest the highest
accuracy of this algorithm.
These findings suggest that machine learning algorithms
can effectively learn about the disease predictions. We may
extend this kind of study to diagnose other diseases. We may
also analyze the past history of data and combine other

American Journal of Computer Science and Technology 2022; 5(3): 146-154 154
machine learning techniques for better study. Other possible
further applications of this study can include such as,
cardiovascular disease prediction, diabetic prediction, breast
cancer prediction, tumor prediction, and multiple disease
predictions.
References
[1] Wikipedia contributors. (2022, June 22). Machine learning. In
Wikipedia, The Free Encyclopedia. Retrieved 06:31, June 26,
2022, from
https://en.wikipedia.org/w/index.php?title=Machine_learning
&oldid=1094363111.
[2] Victor Chang, Vallabhanent Rupa Bhavani, Ariel Qianwen Xu,
MA Hossain. An artificial intellegence model for heart disease
detection using machine learning. Healthcare Analytics,
volume 2, November 2022, 100016.
https://doi.org/10.1016/j.health.2022.100016.
[3] Ghumbre, S. U., & Ghatol, A. A. (2012). Heart disease
diagnosis using machine learning algorithm. In Proceedings of
the International Conference on Information Systems Design
and Intelligent Applications 2012 (INDIA 2012) held in
Visakhapatnam, India, January 2012 (pp. 217-225). Springer,
Berlin, Heidelberg.
[4] Rohit Bharti, Aditya Khamparia, Mohammed Shabaz, Gaurav
Dhiman, Sagar pande, and Parneet Singh. Prediction of Heart
Disease Using a combination of Machine Learning and Deep
learning. Hindawi Computational Intelligence and
Neuroscience, Volume 2021, Article ID 8387680, 11 pages.
https://doi.org/10.1155/2021/8387680.
[5] Khaled Mohamed Almustafa. Prediction of heart disease and
classifiers sensitivity analysis. Almustafa BMC Bioinfirmatics
(2020) 21: 278. https://doi.org/10.1186/s12859-020-03626-y.
[6] Jee S H, Jang Y, Oh D J, Oh B H, Lee S H, Park S W & Yun
Y D (2014), A coronary heart disease prediction model. The
Korean Heart Study. BMJ open, 4 (5), e005025.
[7] Mai Shouman, Tim Turner, and Rob Stocker. Applying k-
Nearest Neighbour in diagnosis heart disease patients..
International Journal of Information and Education
Technology, vol. 2, No. 3, June 2012.
[8] Ganna A, Magnusson P K, Pedersen N L, de Faire U, Reilly
M, Arnlov J & Ingelsson E (2013). Multilocus genetic risk
scores for coronary heart disease prediction. Arteriosclerosis,
thrombosis, and vascular biology, 33 (9), 2267-72.
[9] Raihan M, Mondal S, More A, Sagor M O F, Sikder G,
Majumder M A & Ghosh K (2016, December). Smartphone
based ischeme heart disease (heart attact) risk prediction using
clinical data and data mining approaches, a prototype design.
19
th
International conference on Computer and Information
Technology (ICCIT) (pp. 299-303). IEEE.
[10] Acharya U R, Fujita H, Oh S L, Hagiwara Y, Tan J H &
Adam M (2017). Application of deep convolutional neural
network for automated detection of myocardial infarction
using ECG signals. Information Sciences, 415, 190-8.
[11] Takci H (2018). Improvement of heart attack prediction by the
feature selection methods. Turkish Journal of Electrical
Engineering & Computer Sciences, 26 (1), 1-10.
[12] Brown N, Young T, Gray D, Skene A M & Hampton J R
(1997). Inpatient deaths from acute myocardial infarction,
1982-92: analysis of data in the Nottingham heart attack
register, BMJ, 315 (7101), 159-64.
[13] Soni J, Ansari U, Sharman D & Soni S (2011). Predictive data
mining for medical diagnosis: an overview of heart disease
prediction. International Journal of Computer Applications, 17
(8), 43-8.
[14] Bashir S, Qamar U & Javed M Y (2014, November). An
ensemble-based decision support framework for intelligent
heart disease diagnosis. International Conference on
Information Sociaty (i-Sociaty 2014) (pp. 259-64). IEEE.
[15] Ordonez C (2006). Associate rule discovery with the train and
test approach for heart disease prediction. IEEE Transaction
on Information Technology in Biomedicine, 10 (2), 334-43.
[16] Wikipedia contributors. (2022, June 21). Logistic regression.
In Wikipedia, The Free Encyclopedia. Retrieved 06:36, June
26, 2022, from
https://en.wikipedia.org/w/index.php?title=Logistic_regressio
n&oldid=1094256072.
[17] Wikipedia contributors. (2022, June 1). Linear regression. In
Wikipedia, The Free Encyclopedia. Retrieved 06:39, June 26,
2022, from
https://en.wikipedia.org/w/index.php?title=Linear_regression
&oldid=1091044459.
[18] Wikipedia contributors. (2022, June 4). K-nearest neighbors
algorithm. In Wikipedia, The Free Encyclopedia. Retrieved
06:40, June 26, 2022, from
https://en.wikipedia.org/w/index.php?title=K-
nearest_neighbors_algorithm&oldid=1091525121.
[19] Wikipedia contributors. (2022, June 20). Random forest. In
Wikipedia, The Free Encyclopedia. Retrieved 06:41, June 26,
2022, from
https://en.wikipedia.org/w/index.php?title=Random_forest&ol
did=1094130824.
[20] Wikipedia contributors. (2022, June 15). Decision tree
learning. In Wikipedia, The Free Encyclopedia. Retrieved
06:42, June 26, 2022, from
https://en.wikipedia.org/w/index.php?title=Decision_tree_lear
ning&oldid=1093316444.
[21] Wikipedia contributors. (2022, June 24). Gradient boosting. In
Wikipedia, The Free Encyclopedia. Retrieved 06:43, June 26,
2022, from
https://en.wikipedia.org/w/index.php?title=Gradient_boosting
&oldid=1094845596.
[22] nikki2398. (02 Sep, 2020). ML–Gradient Boosting.
https://www.geeksforgeeks.org/ml-gradient-boosting/.
[23] Wikipedia contributors. (2022, June 20). Support-vector
machine. In Wikipedia, The Free Encyclopedia. Retrieved
06:51, June 26, 2022, from
https://en.wikipedia.org/w/index.php?title=Support-
vector_machine&oldid=1094109362.