Search

The limitations of medical personnel, especially heart disease, cause difficulties in diagnosing heart disorders, so diagnosing heart disorders is not easy, it takes the ability and experience of a cardiologist who has the expertise and experience to be able to accurately diagnose heart disorders. Several studies in the field of computing have been carried out in diagnosing cardiac abnormalities in patients. This study was conducted to accurately test the results of the classification of heart disorders using electrocardiogram medical record data with a C.45 decision tree approach. The results showed that the classification of heart defects obtained a mean squared error (MSE) value of 0.24, a root mean squared error (RMSE) value of 0.49, and an accuracy value of 75.33% with the C4.5 algorithm.

Social media is one of the technological developments that has contributed greatly in making it easy for us to communicate and socialize, one of which is using Twitter social media. Twitter in this study is used as a data source to analyze tweets discussing KRI Nanggala 402. Analysis of KRI Nanggala 402 twitter sentiment is used to see the tendency of public responses to the sinking of the KRI Nanggala 402 submarine whether to give positive or negative opinions. This Sentiment analysis uses the Naïve Bayes Classifier method, which is a classification method. The first research stage is crawling, processing, classification, and evaluation. The classification stage is carried out after the processing phase, where the classification results tend to be positive or negative, using the Naïve Bayes Classifier method. The accuracy of the system in the Sentiment analysis of the KRI Nanggala 402 tweet is 73.00%.

Technological advances at this time are developing very fast, information systems became the frontline in technological advancements, the need for information systems to support jobs is increasingly high. However, its implementation for users does not have a significant impact, so that it needs to be reviewed and re-evaluated in the use of the information system built. The naive bayes classifier method can provide "effective" and "ineffective" conclusions and is used as material for evaluation and improvement. The purpose of this study is to contribute to measuring the effectiveness of the information system, to solve problems with the naïve bayes classifier method approach which has advantages in the process of classifying data and predicting data. From the test results three times, training has been conducted using 100 data, accuracy value of 84.82% and error 15.18%.

Implementation of data mining, machine learning, and statistical data from educational department commonly known as educational data mining. Most of school systems require a teacher to teach a number of students at one time. Exam are regularly being use as a method to measure student’s achievement, which is difficult to understand because examination cannot be done easily. The other hand, programming classes makes source code editing and UNIX commands able to easily detect and store automatically as log-data. Hence, rather that estimating the performance of those student based on this log-data, this study being more focused on detecting them who experienced a difficulty or unable to take programming classes. We propose CLG clustering methods that can predict a risk of being dropped out from school using cluster data for outlier detection.

In the present, data mining can be applied in various fields. One of them is the field of education. By applying data mining in the field of education, education providers can make an analysis of students in their schools. Schools can predict student achievement, make assessments of students more thoroughly, and can also predict students’ interests and talents. This study will discuss the prediction of student learning habits and the prediction of student achievement in order to find out the right steps to take afterwards. In this study, two (2) classification algorithms were used, namely J48 and NaiveBayes. This is done to find the best results from each algorithm.

Using skincare and facial skin care must be in accordance with the type of facial skin, because if it is not suitable, it can cause problems such as facial skin breakouts, dry skin, irritated skin, and acne prone skin. To find out the type of facial skin, you have to do an examination with a skin and genital specialist, but the high cost of consultation and the long queue process is an obstacle for everyone. Therefore, the authors created an expert system to identify facial skin types using a combination of certainty factor methods with forward chaining techniques. The diagnostic results from calculations using an expert system application and the results of manual calculations from one of the respondent data from 100 respondent data, namely producing the same level of confidence, each of which produces a percentage of 99.45% and the diagnostic results state that the user has a normal skin type.

Excess sugar consumption will affect the body's resistance with predicted dangers such as diabetes. Diabetes is a disease that is a quite a dangerous cause of the failure of the pancreas to produce the hormone insulin. Information regarding the achievement of blood sugar targets that must be understood by some people is around 80-130mg / dL because of the lack of effectiveness in consulting directly with experts, the aim of this study is to make a problem solving using an expert system by applying the forward chaining method and certainly a factor. Also implemented on the Android operating system so that the use of every detail of this application can be accelerated by the hand with a smartphone. As a result, every user of this application can ask questions that can diagnose Diabetes Mellitus.Keywords:Expert System, Forward chaining, Diabetes Mellitus Detection, Certainty factor, Android.

In this paper, a new technique of agglomerative hierarchical clustering (AHC), which is known as SLG (single linkage dissimilarity increment distribution, global cumulative score standard), can work well in analyzing students' activity in online learning as evidenced by obtaining the highest score in testing the validity index of cophenetic correlation coefficient (CPCC) ie 0.9237, 0.9015, 0.9967, 0.8853, 0.9875 of the five datasets compared with conventional agglomerative hierarchical clustering methods.