Journal of Artificial Intelligence Computer Engineering Science and Technology

Employee Attrition Prediction Using Logistic Regression and Selectkbest: A Case Study on The IBM HR Analytics Employee Attrition and Performance Dataset

Putra Nurhuda Makatita — 2026-02-15

Employee attrition is a significant strategic concern for organizations as it directly impacts overall performance, productivity, and long-term sustainability. High attrition rates can lead to increased costs in recruitment and training, a loss of skilled and experienced employees, decreased morale among remaining staff, and disruptions to critical business operations. In response to these challenges, many organizations are turning to predictive analytics to anticipate employee turnover and implement effective retention strategies. This study proposes a machine learning-based approach to predict employee attrition using the Logistic Regression algorithm. Logistic Regression is chosen due to its effectiveness in binary classification tasks and its interpretability, which is essential for human resource (HR) professionals when making data-driven decisions. To enhance the model’s performance, the SelectKBest feature selection technique is applied in conjunction with the ANOVA F-test. This method allows the identification of the most influential features contributing to attrition, helping reduce noise and computational complexity while improving model accuracy. The IBM HR Analytics Employee Attrition & Performance Dataset is used in this study. The dataset contains a variety of demographic and organizational attributes such as age, monthly income, job role, tenure, and job satisfaction. The data undergoes a comprehensive preprocessing phase that includes numerical transformation, encoding of categorical variables, normalization, and the implementation of feature selection. By combining Logistic Regression with effective feature selection, this research aims to deliver an accurate and interpretable predictive model. The results are expected to help HR departments proactively identify high-risk employees and take strategic actions to reduce attrition, ultimately supporting better workforce planning and organizational stability.

Analysis of Trends in COVID-19 Mortality and Cure Based on Statistical Data between Provinces Indonesia

Lukman Nur E — 2026-02-15

This study aims to analyze trends in deaths and recoveries from COVID-19 in Indonesia during the period 2020 to 2022 with a quantitative descriptive approach. Data were obtained from the open platform Kaggle and included total cases, recovered cases, and deaths per province each year. The two main epidemiological indicators analyzed were Case Fatality Rate (CFR) and Case Recovery Rate (CRR), which were used to measure the severity and effectiveness of COVID-19 treatment in each region. The analysis was conducted spatially and temporally and visualized using Tableau to highlight inequality between regions. The results show that 2021 is the peak of the pandemic with the highest number of cases and deaths nationally. Provinces on the island of Java, such as East Java, Central Java and DKI Jakarta are under enormous health system pressure, reflected in high CFR and low CRR. In contrast, provinces outside Java such as Papua, West Kalimantan and North Sulawesi show lower CFR and higher CRR, despite having fewer cases. This disparity indicates differences in health facility capacity, population density, effectiveness of data reporting, and speed of medical treatment between regions. Spatial visualization reinforces these findings by showing that areas with high mobility and dense urbanization tend to be centers of virus spread. This study emphasizes the importance of a region-based pandemic response strategy, tailored to local characteristics. Recommendations include strengthening health infrastructure, improving the quality of reporting and data recording systems, and developing data-based responsive policies. The results of this study are expected to serve as a reference in the formulation of more adaptive and equitable public health policies, as well as a basis for further research that is more in-depth with a multidimensional approach.

Financial Literacy and Lifestyle Analysis on Interest in Using Online Loans Among Students

Sehan Agus Prasetio — 2026-02-15

The development of technology in the financial sector, especially online loan services, is now an alternative option that many students look at to meet their financial needs. Unfortunately, there are still many students who do not have a good understanding of financial management, coupled with a consumptive lifestyle that can have negative impacts, such as being entangled in debt, disrupted in the lecture process, and mental health problems. This research was conducted using the Systematic Literature Review (SLR) method through a study of a number of scientific articles published on Google Scholar during the 2020–2024 period. The goal is to find out how financial literacy and lifestyle affect students' interest in using online loans. The results of the study show that students' ability to understand finance has a large and significant effect on their interest in using this service. On the contrary, lifestyle has not been shown to have a significant influence. Therefore, increasing financial literacy among students is essential so that they can make wise financial decisions, especially when it comes to utilizing online loan services.

Electric Current Prediction Based on Voltage and Frequency Using K-Nearest Neighbors (KNN) and Ridge Regression Algorithms

Silvia Kurnia — 2026-02-15

The development of information technology has had a significant impact on various aspects of life, including the field of electricity. Technology, derived from the words technologia and techno, means expertise and knowledge. As time progresses, many tasks that were previously performed manually can now be simplified and automated through the use of digital technologies, including in the calculation of electric current. Calculating current based on frequency and voltage values involves a high level of complexity when done manually. To address this challenge, this study designs a system for predicting electric current using a Machine Learning approach, specifically employing the K-Nearest Neighbors (KNN) and Ridge Regression algorithms. KNN is an instance-based learning method that classifies or predicts new data based on its proximity to existing data. Meanwhile, Ridge Regression is a form of linear regularization that is effective in handling multicollinearity and overfitting in regression modeling. The methodology includes several stages: data collection, normalization, splitting the data into training and testing sets, model training, and performance evaluation using metrics such as Mean Squared Error (MSE) and the coefficient of determination (R²). The model is developed using Python as the primary programming language due to its capabilities in supporting data analysis and implementing machine learning algorithms efficiently. The results of the study show that the designed system is capable of effectively predicting electric current based on frequency and voltage inputs. Additionally, a performance comparison between the KNN and Ridge Regression algorithms is obtained, which can serve as a consideration in selecting the most appropriate prediction method based on the characteristics of electrical data. These findings are expected to contribute to the utilization of intelligent technology for more efficient and effective technical analysis in the future.

Keywords: information technology, KNN, Ridge Regression, Machine Learning, research methodology.

Cooling Hot Coffee Using Nonlinear Equation Numerical Methods

Fahrij Arrohman — 2026-02-15

Freshly brewed hot coffee requires time to reach the ideal temperature for comfortable consumption, typically around 58–66 °C. Consumers generally rely on estimation or experience when waiting, without a precise scientific approach. This study examines the cooling process of coffee as a heat transfer phenomenon that can be modeled using Newton’s Law of Cooling, in which the rate of temperature decrease is proportional to the temperature difference between the object and its environment. Since the governing equation is nonlinear, solving it requires a numerical method. The Newton-Raphson method was selected for its efficiency in solving single-variable nonlinear equations and its fast convergence.

The simulation was conducted using Python, with the following parameters: an initial temperature of 80.57 °C, ambient temperature of 27 °C, target consumption temperature of 62.99 °C, and a cooling constant of 22.24×10⁻³ s⁻¹ based on previous experimental data. The results showed that the ideal consumption temperature is reached in approximately 17.88 minutes. The iteration graph demonstrated a rapid decrease in function values, requiring only four iterations to converge. While the simulation showed high accuracy during the initial cooling phase, minor deviations occurred as the temperature approached ambient levels. This discrepancy is likely due to the assumption of a constant cooling coefficient, whereas in reality it may vary depending on the solution temperature and air convection conditions.

This model can be used to objectively predict the optimal time to enjoy coffee based on scientific calculations. In addition to benefiting home consumers seeking the best coffee-drinking experience, the findings of this research can be applied in the culinary and hospitality industries to consistently and efficiently serve hot beverages at the right temperature. By providing predicted consumption times, the risk of burns from excessively hot drinks can be minimized, thereby enhancing customer satisfaction. This approach can also be adapted for other hot beverages requiring temperature control, offering broader applicability in the fields of food engineering and thermal design.