A Machine Learning Framework for Improving Classification Performance on Credit Approval

Credit scoring is a model commonly used in the decision-making process to refuse or accept loan requests. The credit score model depends on the type of loan or credit and is complemented by various credit factors. At present, there is no accurate model for determining which creditors are eligible for loans. Therefore, an accurate and automatic model is needed to make it easier for banks to determine appropriate creditors. To address the problem, we propose a new approach using the combination of a machine learning algorithm (Naïve Bayes), Information Gain (IG), and discretization in classifying creditors. This research work employed an experimental method using the Weka application. Australian Credit Approval data was used as a dataset, which contains 690 instances of data. In this study, Information Gain is employed as a feature selection to select relevant features so that the Naïve Bayes algorithm can work optimally. The confusion matrix is used as an evaluator and 10-fold cross-validation as a validator. Based on experimental results, our proposed method could improve the classification performance, which reached the highest performance in average accuracy, precision, recall, and f-measure with the value of 86.29%, 86.33%, 86.29%, 86.30%, and 91.52%, respectively. Besides, the proposed method also obtains 91.52% of the ROC area. It indicates that our proposed method can be classified as an excellent classification.

Analysis of Conti Ransomware Attack on Computer Network with Live Forensic Method

Ransomware viruses have become a dangerous threat increasing rapidly in recent years. One of the variants is Conti ransomware that can spread infection and encrypt data simultaneously. Attacks become a severe threat and damage the system, namely by encrypting data on the victim's computer, spreading it to other computers on the same computer network, and demanding a ransom. The working principle of this Ransomware acts by utilizing Registry Query, which covers all forms of behavior in accessing, deleting, creating, manipulating data, and communicating with C2 (Command and Control) servers. This study analyzes the Conti virus attack through a network forensic process based on network behavior logs. The research process consists of three stages, the first stage is simulating attacks on the host computer, the second stage is carrying network forensics by using live forensics methods, and the third stage is analysing malware by using statistical and dynamic analysis. The results of this study provide forensic data and virus behavior when running on RAM and computer networks so that the data obtained makes it possible to identify ransomware traffic on the network and deal with zero-day, especially ransomware threats. It is possible to do so because the analysis is an initial step in generating virus signatures based on network indicators.

The DHCP Snooping and DHCP Alert Method in Securing DHCP Server from DHCP Rogue Attack

DHCP Server as part of the network infrastructure in charge of distributing host configurations to all devices has the potential to be controlled. If the DHCP Server is successfully controlled, all network devices connected to the server can potentially be controlled. From the observations made at PT. Rekayasa Engineering found a vulnerability in the DHCP Server that has the potential to experience DHCP Rogue or DHCP Spoofing, where the client will fail to communicate with the authorized DHCP Server, as well as open the door for attackers to enter the network. For this reason, DHCP Snooping and DHCP Alert methods are implemented. DHCP Snooping will ensure that every data traffic has been filtered and directed to the registered interface. Meanwhile, the use of DHCP Alert is required in monitoring data traffic during the Discover, Offer, Request, and Acknowledge (DORA) process. In the tests performed, DHCP Snooping and DHCP Alert managed to anticipate attacks that tried to placed DHCP Rogue on the network infrastructure. DHCP Alert, configured on the proxy router, ensures that the DORA process can only occur between an authorized DHCP server and a client. DHCP Snooping test also shows that communication from clients can only be replied to by Trusted DHCP Server. The existence of DHCP Snooping and DHCP Alert makes the host configuration fully controlled by the authorized DHCP Server.

Information System for Monitoring High-Risk Pregnant Women

Maternal mortality rates are still high in several areas, including Bantul Regency, Special Region of Yogyakarta. Based on the data obtained from the Bantul District Health Office, from 2018 to 2019, 28 pregnant women died. Posyandu and Puskesmas cadres often encounter problems in collecting data on pregnant women because they still use manual methods. Manual records using books has disadvantages because sometimes pregnant women forget to bring their books. Therefore, an application for recording pregnancy history is needed to enable convenient monitoring by Posyandu cadres, Puskesmas, doctors, and hospitals in order that pregnant women patients can be handled properly in case of emergency. The application used by pregnant women is Mobile App-based, meanwhile, the Web-based Monitoring Information System is used by Posyandu cadres, Puskesmas, doctors, and hospitals. The application allows displaying the medical history and makes it easier for pregnant women to have counseling or examinations without meeting directly with the doctor. If there are any problems in the womb, the doctor will immediately provide a solution or recommendation. The results of the system testing with 15 respondents as users show that 52,1% strongly agree, 37,7% agree, and 10,2% neutral in response to the system interface. The implementation of the information system for monitoring high-risk pregnant women in the majority is accepted by all actors. Hence, it can be concluded that in an attempt to digitalize manual recording of pregnant women's examinations, this information system for monitoring high-risk pregnant women is reliable to be implemented.

Sentiment Analysis and Topic Modeling of Indonesian Public Conversation about COVID-19 Epidemics on Twitter

The World Health Organization (WHO) declared the COVID-19 outbreak has resulted in more than six million confirmed cases and more than 371,000 deaths globally on June 1, 2020. The incident sparked a flood of scientific research to help society deal with the virus, both inside and outside the medical domain. Research related to public health analysis and public conversations about the spread of COVID-19 on social media is one of the highlights of researchers in the world. People can analyze information from social media as supporting data about public health. Analyzing public conversations will help the relevant authorities understand public opinion and information gaps between them and the public, helping them develop appropriate emergency response strategies to address existing problems in the community during the pandemic and provide information on the population's emotions in different contexts. However, research related to the analysis of public health and public conversations was so far conducted only through supervised analysis of textual data. In this study, we aim to analyze specifically the sentiment and topic modeling of Indonesian public conversations about the COVID-19 on Twitter using the NLP technique. We applied some methods to analyze the sentiment to obtain the best classification method. In this study, the topic modeling was carried out unsupervised using Latent Dirichlet Allocation (LDA). The results of this study reveal that the most frequently discussed topic related to the COVID-19 pandemic is economic issues.

Song Recommendation Application Using Speech Emotion Recognition

This research was conducted to facilitate the interaction between radio broadcasters and radio listeners during the song request process. This research was triggered by the difficulty of the broadcasters in monitoring song requests from listeners. The system is made to accommodate all song requests by listeners. The application produced in this study uses speech emotion recognition technology based on a person's mood obtained from the spoken words. This technology can change the voice into one of the mood categories: neutral, angry, sad, and afraid. The k-Nearest Neighbor method is used to get recommendations for recommended song titles by looking for the closeness of the value between the listener's mood and the availability of song playlists. kNN is used because this method is suitable for user-based collaborative problems. kNN will recommend three songs which then be offered to listeners by broadcasters. Based on tests conducted to the broadcasters and radio listeners, this study has produced a song request application by recommending song titles according to the listener's mood, the text message, the searching songs, and the song requests and the song details that have been requested. Functional test that has been carried out has received 100 because all test components have succeeded as expected.

Self-Evaluation of Jurnal Informatika Sunan Kalijaga (JISKa): Perspectives of Reviewers and Authors

The success of JISKa is inseparable from the role of reviewers and authors. Unfortunately, JISKa had never been assessed or evaluated by reviewers and authors despite the fact that assessment from the reviewers and authors would be valuable feedback for JISKa’s self-evaluation. Therefore, survey-based research has recently been conducted to assess JISKa’s performance using the User Acceptance Test of OJS version 2.4.8.0. This study used a survey method to obtain an assessment and evaluation from reviewers and authors related to JISKa.The respondents in this study consist of 68 authors and 26 reviewers. The result of this study stated that 91.2% of the authors and 84.6% of reviewers are satisfied with JISKa. A percentage number of 100% of writers and reviewers wants JISKa to raise its level of Sinta accreditation. This accreditation is awarded in 2018 and will end in 2023. JISKa is now on Sinta 4.The JISKa website appearance looks good and easy to use. The dashboard on the JISKa page is user-friendly for the author. However, the current version of JISKa OJS 2.4.8.0 needs to be upgraded to OJS version 3. There are some points for the future consideration of JISKa: JISKa needs to promote itself more, upgrade the OJS version, and provide the reviewers with certificates of appreciation for future consideration.

The Social Engagement to Agricultural Issues using Social Network Analysis

Twitter is one of the micro-blogging social media which emphasizes the speed of communication. In the 4.0 era, the government also promotes the distribution of information through social media to reach the community from various lines. In previous research, Social Network Analysis was used to see the relationship between actors in a work environment, or as a basis for identifying the application of technology adoption in decision making, whereas no one has used SNA to see trends in people's response to agricultural information. This study aims to see the extent to which information about agriculture reaches the community, as well as to see the community's response to take part in agricultural development. This article also shows the actors who took part in disseminating information. Data was taken on November 13 to 20, 2020 from the Drone Emprit Academic, and was taken limited to 3000 nodes. Then, the measurements of the SNA are represented on the values of Degree Centrality, Betweenness Centrality, Closeness Centrality, and Eigenvector Centrality. @AdrianiLaksmi has the highest value in Eigenvector Centrality and Degree Centrality, he has the greatest role in disseminating information and has many followers among other accounts that spread the same information. While the @RamliRizal account ranks the highest in Betweenness Centrality, who has the most frequently referred information, and the highest Closeness Centrality is owned by the @baigmac account because of the fastest to re-tweet the first information.

Research Trend of Causal Machine Learning Method: A Literature Review

Machine learning is commonly used to predict and implement pattern recognition and the relationship between variables. Causal machine learning combines approaches for analyzing the causal impact of intervention on the result, asumming a considerably ambigous variables. The combination technique of causality and machine learning is adequate for predicting and understanding the cause and effect of the results. The aim of this study is a systematic review to identify which causal machine learning approaches are generally used. This paper focuses on what data characteristics are applied to causal machine learning research and how to assess the output of algorithms used in the context of causal machine learning research. The review paper analyzes 20 papers with various approaches. This study categorizes data characteristics based on the type of data, attribute value, and the data dimension. The Bayesian Network (BN) commonly used in the context of causality. Meanwhile, the propensity score is the most extensively used in causality research. The variable value will affect algorithm performance. This review can be as a guide in the selection of a causal machine learning system.

Online Integrated Development Environment (IDE) in Supporting Computer Programming Learning Process during COVID-19 Pandemic: A Comparative Analysis

COVID-19 has spread to various countries and affected many sectors, including education. New challenges arise in universities with study programs related to computer programming, which require a lot of practice. Difficulties encountered when students should setting up the environment needed to carry out programming practices. Furthermore, they should install a text editor called Integrated Development Environment (IDE) to support it. There is various online IDE that supports computer programming. However, students must have an internet connection to use it. After all, many students cannot afford to buy internet quotas to access online learning material during the COVID-19 pandemic. According to these problems, this study compares several online IDEs based on internet data usage and the necessary supporting libraries' availability. In this study, we only compared eleven online IDEs that support the Python programming language, free to access, and do not require logging in. Based on the comparative analysis, three online IDEs have most libraries supported. They are REPL.IT, CODECHEF, and IDEONE. Based on internet data usage, REPL.IT is an online IDE that requires the least transferred data. Moreover, this online IDE also has a user-friendly interface to place the left and right sides' code and output positions. It prevents the user from scrolling to see the results of the code that has been executed. The absence of advertisements also makes this online IDE a more focused appearance. Therefore, REPL.IT is highly recommended for users who have a limited internet quota, primarily to support the learning phase of computer programming during the COVID-19 pandemic.