JOURNAL TECH-E - VOL.
4 NO.
Online Version Available at: http://bsti.
id/e-jurnal JOURNAL OF TECH-E .
1-1916 (Onlin.
| 2598-7585 (Printe.
| Article Comparison of Data Mining Methods Using C4.
Algorithm and Naive Bayes in Predicting Heart Disease Rino Buddhi Dharma University.
Faculty of Sains & Technology.
Banten.
Indonesia
SUBMISSION TRACK
A B S T R A C T
Recieved: Februari 19, 2021 Final Revision: Februari 26, 2021 Available Online: March 15, 2021 Heart disease is a condition of the presence of fatty deposits in the coronary arteries in the heart which changes the role and shape of the arteries so that blood flow to the heart is Data mining methods can predict this disease, some of the methods are C4.
5 Algorithm and Naive Bayes which are often used in research.
The data set in this research was obtained from the uci machine learning repository site, where the dataset has 3546 records and 13 attributes.
The accuracy value of the Nayve Bayes algorithm has a high value of 81.
40% compared to the C4.
5 algorithm which only has an accuracy value of 79.
Based on the calculation results, it can be concluded that the Nayve Bayes Algorithm is a very good clarification because it has a value between 709 - 1.
From conclusion above, the Nayve Bayes algorithm has a higher accuracy value than the C4.
5 algorithm so the researchers decided to use the Nayve Bayes algorithm in predicting heart disease.
KEYWORD
Heart Disease.
Data Mining.
C4.
Nayve Bayes CORRESPONDENCE E-mail: rino@ubd.
INTRODUCTION
Heart disease is a condition of the presence of fatty deposits in the coronary arteries in the heart which changes the role and shape of the arteries so that blood flow to the heart is delayed .
The World Health Organization in 2013 states that the death rate caused by heart disease is 45%, and it is estimated that in 2030 it will increase by 23.
3 million people each year.
In 2013, the prevalence of heart disease diagnosed by doctors in Indonesia was 883,447 people.
Central Java Province is in third place with a total of 120,447 heart disease sufferers .
asic health Heart disease occurs indirectly, usually a person will experience a process of narrowing of the coronary vessels in a long enough time, therefore everyone has a risk of heart disease.
In addition, there are other factors that cause a person to experience heart disease, namely lifestyle and genetic Currently, technological developments have entered many fields, one of which is in the medical world.
The use of information technology (IT) in the medical world is familiar to the wider community and one of these uses is used to predict heart disease.
There are many ways or methods to predict RINO / J OURNAL TECH-E - VOL.
NO.
heart disease, one of which is using data Data mining (DM) is a combination of a number of computer science disciplines .
, which defines it as the process of discovering new patterns from very large data sets, including methods that are a slice of artificial intelligence, machine learning, statistics, and database systems .
Data mining or word addition is a relatively fast and easy technique to find patterns and / or relationships between data, automatically.
By combining four computer science disciplines as defined above, knowledge can be found in five sequential processes:
selection, processing, transformation, data mining, and interpretation / evaluation .
Data mining is a data processing method where does it work to find hidden patterns from some dataset and the results its can be used to make decisions in the future.
This data mining is also known as pattern recognition .
Data mining is a large-scale data processing method, therefore data mining has an important role in industry, finance, weather, science and technology.
In general, data mining studies discuss methods such as clustering, classification, regression, variable selection and market basket analysis.
From the above definitions, it can be concluded that in general data mining is a data analysis activity to look for a certain pattern, with a large amount of data and aims to produce information that can be used and developed The C4.
5 algorithm is a program that contributes to a data set labeled and produces a decision tree as output .
This follow-up decision tree is then verified against invisible labeled test data to calculate generalizations.
C4.
5 is a program used to generate taxonomic rules using a decision tree from a given data set.
The C4.
5 algorithm is an extension of the basic ID3 algorithm and was designed by Quinlan.
C4.
5 is one of the most widely used learning algorithms.
The C4.
algorithm constructs a decision tree from a series of training data similar to the ID3 algorithm, using the information entropy C4.
5 is also known as statistical One of the classification algorithms that is often used and has received a lot of attention from researchers in predicting heart disease is Naive Bayes and Decision Tree (C.
Simplicity of the Naive Bayes algorithm and the Decision Tree (C4.
What makes these two algorithms attractive are because their have high accuracy for prediction and can be implemented in various applications, such as expert system, data mining for prediction and classification .
Because of this, we compare the two algorithms from their level of accuracy and time performance to improve prediction performance.
Naive Bayes is one of the algorithms contained in the clarification technique.
Naive Bayes is a clarification with the probability and statistical method proposed by the British scientist Thomas Bayes, namely predicting future opportunities based on previous experiences so it is known as the Bayes Theorem.
The theorem is combined with Naive where it is assumed that the conditions between the Clarification Naive Bayes assumed that the presence or absence of certain characteristics of a class has nothing to do with LITERATURES REVIEW The definition of the C4.
5 algorithm was introduced by Quinlan .
as an improved version of ID3 .
In ID3, decision tree induction can only be done for categorical features .
ominal or ordina.
, while numeric types .
nterval or rati.
cannot be used.
Improvements made are being able to handle features with numeric types, prunning decision trees, and deriving rule Bayes is a simple probability-based prediction technique that differs from the application of the Bayes theorem .
r Bayes rul.
with the assumption of strong .
In other words, in Nayve Bayes, the model used is an independent feature model.
RINO / J OURNAL TECH-E - VOL.
NO.
In Bayes .
specially nayve Baye.
, the meaning of strong independence on features is that a feature in a data set is not related to the presence or absence of other features in the same data.
For example in the case of animal clarification with features of skin Model A model is a class that underlies the process logic in a software application and the classes associated with it.
Model is an object that does not contain information about the user interface.
Model is also a class that contains methods / functions and is used to store data and relevant business rules.
II.
FRAMEWORK
Model-View-Controller (MVC) programming model that implements an application architecture into three parts, namely separating the process, views and parts that connect to the database.
MVC aims to separate business processes from user interface considerations so that developers can more easily develop one part of the application so that it does not affect other parts .
In the MVC model describes information .
and business processes.
The view contains interface elements such as text, images, or 16 input forms, while the controller manages the communication between the view and the model .
mapped an MVC workflow will look like the following picture.
Figure 1: MVC Workflow The MVC concept describes the ModelView-Controller as follows .
View The view view is a collection of classes that represent the elements in the interface, in the view there are names that are used to identify 17 view script files when called via the render function.
The view name is the same as the view script file name.
Controller Controller is a class that connects the model and view, used to communicate between classes in the model and view.
Controllers have standard actions.
When the user request does not specify which action to run, the program executes the standard action.
Expert system Understanding Expert Systems According .
, some definitions of expert systems according to some experts are as follows.
According to Durkin: An expert system is a computer program designed to model the problemsolving abilities of an expert.
According to Ignizo: Expert system is a model and related procedures, in a particular domain, where the level of expertise can be compared with the expertise of an expert.
According to Giarratano and Riley:
An expert system is a computer system that can match or imitate the abilities of an expert RINO / J OURNAL TECH-E - VOL.
NO.
METHODS
The dataset to be studied is obtained from the uci machine learning repository data https://archive.
edu/ml/datasets/Heart Disease where the dataset has 3546 records, 13 attributes of these attributes are the result Table 1.
Dataset From UCI Website Sex
Age
CSk
CPD
PHyp
Dbts
TCh
SBP
DBP
BMI
TYCHD
26,97
28,73
25,34
28,58
33,11
21,68
26,36
23,61
22,91
27,64
Algorithm C4.
The following is the model used in processing the dataset using the C4.
Naive Bayes The following is a model used in dataset processing using the Naive Bayes algorithm:
Figure 4: Naive Bayes Algorithm processing model in Rapid Miner Studio Figure 2: Dataset Processing Model Figure 3: C4.
5 Algorithm Processing Process in Rapid Miner Studio Figure 5: Naive Bayes Algorithm Model Processing in Rapid Miner Studio RINO / J OURNAL TECH-E - VOL.
NO.
The use case of an expert information system using the nayve Bayes algorithm is as People who regularly consume the smallest cigarette even though it's only one cigarette a Table 4.
Table Description of Active Smokers Value Information Passive Smoker Active Smoker A day's cigarette stick People who smoke a certain amount in one Table 5.
Table of Number of Cigarettes in a Day Range Information Figure 6: Use case diagram of a heart disease expert system Value 4: without symptoms Value 3: non-angina pain IV.
RESULT
For the diagnosis of heart disease, there are 13 criteria used, which are as follows.
Value 2: angina atipikal Value 1: angina tipikal Age of the patient The patient age ranges in the table below are a reference for heart disease.
The usual stroke People who have a stroke Table 6.
Stroke Information Table Table 2.
Age Range Table in Heart Disease Range Information Young Middle-Aged Old Gender The sexes generally recognized throughout the world are male and female.
Table 3.
Table Description of Gender Value Woman Man Active Smoker Information No Value Information Not Have stroke Have stroke Sugar levels (> 120 mg / d.
Sugar level is the level of glucose in the The unit used for sugar content is mg / dl .
illigrams per decilite.
Examination of blood sugar levels in venous blood when the patient is fasting 12 hours before the examination .
lood sugar satisfied nuchte.
or 2 hours after eating .
ost prandial blood Table 7.
Information Table of Blood Sugar Value Information Yes RINO / J OURNAL TECH-E - VOL.
NO.
Diabetes People who have diabetes Middle High Table 8.
Information Table Diabetes Value Information Yes Cholesterol People who have high cholesterol have a major role in triggering the occurrence of blockage of blood vessels that leads to heart attacks, strokes and others.
Table 9.
Cholesterol Information Table Value Information Low Middle High Pressure when the Heart Pumps the Whole Body The pressure in the blood vessels when the heart contracts to pump clean blood around the body.
Table 10.
Table Description of Pressure when the Heart Pumps Throughout the Body Range Body Weight BMI This is calculated by dividing the body weight in kilograms by the height in meters Suppose you have a weight of 75 kg and a height of 1.
Your Body Mass Index or BMI is: BMI = 75kg / .
= 27.
Table 12.
Table Information on Body Weight BMI Range Thin Ideal Fat Heartbeat Heartbeat is the pulse released by the heart and as a result of blood flow through the Doctors usually use a stethoscope when examining a patient to listen for a Table 13.
Table Description of Heart Rate Range Slow Normal Fast Information Low Middle High Blood Pressure when the Heart Relaxes Blood pressure when the heart is relaxing or On a heart rate curve, diastolic pressure is the blood pressure represented in the range between the heart rate charts.
Range Table 14.
Table Description of Diagnosis of Heart Narrowing Information Low Information Diagnosis of heart disease Narrowing and hardening of the arteries due to plaque buildup on the walls of blood This condition is a common cause of heart disease.
No Range Table 11.
Table Description of Blood Pressure when the Heart Relaxes Information Information <50% Narrowing of the Narrowing of the diameter RINO / J OURNAL TECH-E - VOL.
NO.
DISCUSSION