APPLICATION AND ATTRIBUTE ANALYSIS IN THE MODEL OF CLASSIFYING HEART DISEASE

The heart is the central center in the human circulatory system. A malfunction of the heart that is not functioning is a condition in which the heart cannot carry out its duties properly. Selection of features that can reduce a very large dataset and in a data set that is not suitable can use a reduction model. The classification process is strongly influenced by an attribute. Various types of inappropriate redundancy have a negative effect on classification accuracy. Heart disease data was taken from the UCI Machine Learning Repository dataset. In this study, the researchers used the K-Nearest Neighbor (KNN) algorithm where the K-Nearest Neighbor algorithm can classify the results of heart disease accurately. The results are as follows 1.67358 rank one 1.33949 rank two, 1.27260 rank three, 1.2528 rank four, 1.24193 rank last.


Introduction
Cardiovascular disease is a disease that is often caused by blood vessels and heart function in humans.Attribute features are very important in the accuracy of the results, so it is necessary to know the main attributes of a disease [1].Different results are often obtained when diagnosing heart disease, so important attribute features are selected based on the efficiency of diagnosis [2].
Diagnosis of different diseases is usually carried out in making a diagnosis decision to the examination of the patient and also the experience of the doctor.A new technology is needed to be developed on a new model of monitoring system technology that uses rfid technology [3].For heart resistance, minerals and vitamins play an important role for the human body.Other causes of vitamin and mineral deficiencies can be determined earlier by using the field of artificial intelligence through expert systems [4].In the examination carried out, not all attribute conditions are met, but the diagnostic process must still be carried out.
The classification process is influenced by attributes, then after redundancy is carried out on the attributes, it does not reduce the level of accuracy of the dataset [5].Attribute reduction has become an important step in pattern recognition and machine learning tasks [6].This inspired the author to add an attribute reduction process in making classification decisions on the heart disease dataset.K-Nearest Neighbor (KNN) belongs to the instance-based learning group.This algorithm is also a lazy learning technique.There are many ways to measure the proximity between new data and old data (training data), including the Euclidean distance and Manhattan distance (city block distance), the most commonly used is the Euclidean distance [7].A high score indicates that the feature and the target class cannot be independent and therefore we have to store the feature in a new dataset [8].K-Nearest Neigbhor algorithm to obtain an accurate and effective level of accuracy without reducing the integrity of the original data reduction techniques while maintaining the integrity of the original data.So the reduced data set must be more efficient and produce an analysis that is close to each other [9].K-Nearest Neighbor (KNN) belongs to the instance-based learning group.The KNN algorithm is done by looking for objects in the training data that are close to the objects in the new data or testing data [7] K-Nearest Neighbor can be defined as a classifer that is used to classify a data with a comparison of the K values in the nearest neighbor, the K parameter in the K-Nearest Neighbor has a very important influence on the final prediction results obtained.(, ) = √ ∑ =1  (  −   ) 2…………………………. (3)  The calculation of the level of accuracy in confusion matriks:

Method
Data collection is done by taking data sets of heart disease obtained from the University of California Irvine (UCI) Machine Learning Repository which is public which will be divided into training and testing data.Problem Analysis At this stage an analysis of the analysis of attribute reduction in classifying heart disease is carried out.The problem that arises is that the increasing number of relevant attributes will affect execution time, and can increase the accuracy of the algorithm so attribute reduction is needed.generated information.Design implementation in this research algorithm is K-Nearest Neighbor (KNN) to get accuracy results from both algorithms in terms of reducing complex data without reducing the integrity of the original data and not reducing the quality of information generated from the data.Testing of the heart disease dataset using the K-Nearest Neighbor (KNN) algorithm.Methods of Data Analysis using quantitative data, namely data related to numbers or quantities consisting of 13 attributes, namely age, gender, type of chest pain, blood pressure, cholesterol, sugar levels, electrocardiography, blood pressure, induced angina, oldpeak, segment_st, flaurosopy , and heart rate.

Results and Discussion
The K-Nearest Neighbor model used has an influence on the final prediction results obtained with the nearest neighbor value.In the research model using the KNN model, evaluation is carried out to determine the level of accuracy of the results of using preprocessed data, attribute reduction, classification of test data .The following is to see the values of min, max, new min, new max, mean Std are as follows:

PRELIMINARY DATA age sex cp trestbps chol fbs restecg thalach exang oldpeak slope ca thal target
The next step is to see the value of 2 data types, namely integer and real data, so it is necessary to normalize the data using the Min-Max method, following the normalized dataset with the formula as below.
The following is training data to be able to see the value in classifying heart disease datasets as below: Furthermore, for testing data, it can be seen in classifying heart disease datasets as below: in looking at the Euclidean Distance value in the ranking of the testing data samples, the KNN process will be carried out by seeing whether the dataset is in the row, then after carrying out the KNN process, accurate or inaccurate results will be obtained according to the existing data by looking for a value against Y using the Euclidean formula.classification class in value and each attribute value normalized with the KNN model is as follows: The final results obtained from calculations using the KNN model can be visualized with a graph as shown below: The value of K is obtained from several tests, so that the value of K = 10 with the highest accuracy, so for taking the value of accuracy in this study using K = 10.It can be concluded that based on the graphic above, using the calculation of the KNN model, an accuracy of 84% is obtained.2. Based on the results of the test to reduce the 4th attribute (SkinThickness/Skin Thickness) the accuracy of KNN before it was reduced was 82.71%.3.By obtaining the results of KNN and Chi-Square, the results obtained by the Chi-Square method are much more accurate than the KNN method, but the Chi-Square method is more general in classifying the data to obtain high accuracy results. .

ACKNOWLEDGEMENT
This research was conducted in collaboration with the Department of Health and Cut Meutiadan Hospital.It is hoped that this research can contribute to universities and other researchers as a reference and further development.Thanks to all who participated in completing this research.

Figure 2 :
Figure 2 : Classification of Each Attribute

Figure 3 .
Figure 3. Final Result of KNN Model Calculation 4. RESULT 1.The value of K is obtained from several tests, so that the value of K = 10 with the highest accuracy, so for taking the value of accuracy in this study using K = 10.It can be concluded that based on the graphic above, using the calculation of the KNN model, an accuracy of 84% is obtained.2. Based on the results of the test to reduce the 4th attribute (SkinThickness/Skin Thickness) the accuracy of KNN before it was reduced was 82.71%.3.By obtaining the results of KNN and Chi-Square, the results obtained by the Chi-Square method are much more accurate than the KNN method, but the Chi-Square method is more general in classifying the data to obtain high accuracy results..5.ACKNOWLEDGEMENTThis research was conducted in collaboration with the Department of Health and Cut Meutiadan Hospital.It is hoped that this research can contribute to universities and other researchers as a reference and further development.Thanks to all who participated in completing this research.