International Journal of Recent Technology and Applied Science (IJORTAS)

A Low-Cost Arduino-Based Smart Parking Assistance System Using Distance Sensors for Real-Time Vehicle Guidance

Harfian Rais — 2026-09-07

The growing number of vehicles has made parking increasingly challenging, particularly in areas with limited parking space. Conventional parking methods often rely on drivers' visual judgment, which may lead to inaccurate distance estimation and increase the risk of minor collisions. To address this issue, this study presents the design and implementation of a low-cost smart parking assistance system based on an Arduino Uno microcontroller integrated with an ultrasonic distance sensor and LED indicators that provide real-time visual guidance. The system was developed using the Arduino IDE and experimentally evaluated by comparing sensor measurements with actual distances ranging from 10 to 60 cm, while also examining the accuracy of distance detection and the reliability of LED responses. The experimental results showed stable and consistent performance, with a maximum measurement error of only ±1 cm across all test distances. In addition, the system achieved a 100% success rate in LED indicator activation, where the green, yellow, and red LEDs responded accurately according to the predefined safety thresholds without noticeable delay. These findings demonstrate that the proposed system can provide reliable real-time obstacle detection, helping drivers park more safely while reducing the risk of collisions. Owing to its simple design, low implementation cost, and dependable performance, the proposed system is well suited for residential garages, university campuses, and small commercial parking facilities. Future research should explore the integration of multi-sensor fusion, Internet of Things (IoT)-based remote monitoring, cloud-based data analytics, and artificial intelligence techniques to enable predictive parking occupancy analysis and intelligent parking management for large-scale smart city applications.

Comparative Evaluation and Kinetics of Active, Passive, and Hybrid Treatment for Acid Mine Drainage Neutralization and Fe–Mn Precipitation Using Goat Manure as Organic Substrate

Marwita — 2026-09-15

Acid Mine Drainage (AMD) from sulfide-mineral oxidation poses an urgent threat to receiving water bodies due to its low pH and dissolved-metal load, making efficient, sustainable neutralization a priority for small- to medium-scale mines. This study comparatively evaluated active treatment (Ca(OH)₂), passive treatment (goat manure as an SRB substrate), and a hybrid combination for AMD neutralization and Fe–Mn removal, using a 15-day laboratory batch experiment monitoring pH, TSS, Fe, and Mn against an untreated control, analyzed via effectiveness calculations and Two-Way ANOVA. All treatments raised pH toward neutral, with passive treatment achieving the highest effectiveness (75% pH improvement, 52% Fe removal, 67% Mn removal); pH improvement plateaued near this level once added alkalinity reached equilibrium with residual acidity from ongoing pyrite oxidation. TSS rose in all treatments most in the hybrid system because effective metal removal itself generates suspended Fe/Mn-hydroxide flocs and SRB biomass, though values stayed within the regulatory limit (400 mg/L). These results position goat manure-based passive and hybrid treatments as low-cost, mechanistically distinct alternatives for net-acid, low-metal AMD management. Given the single-reactor, non-replicated design, further research with biological replicates, controlled temperature/DO/ORP monitoring, and field-scale validation is recommended, particularly to optimize passive-system design for TSS mitigation.

Interpretable Machine Learning Framework for Personalized Health Insurance Risk Prediction

Mohammed Al-Mhadawi — 2026-09-07

In modern health insurance systems, accurate medical expenditure prediction is vital for financial solvency and equitable premium distribution. However, deployment is often hampered by the trade-off between predictive accuracy and model transparency. This study presents a unified, interpretable machine learning regression framework to evaluate personalized health insurance charges using structured tabular data. We benchmarked six predictive architectures (five tree ensembles and a multi-layer perceptron control) on a real-world dataset (n=1,338). Experimental results demonstrate that Gradient Boosting achieved superior performance with R²=0.8789, RMSE = $4,335.47, MAPE = 28.49%, and an operational model footprint of only 170 KB. In contrast, standard deep learning (MLP) failed catastrophically (R²= −0.3947) due to severe right-skewness and nonlinear tabular feature interactions. Model interpretability via SHAP values identified smoker status (48.2% Gini importance) and BMI interactions as primary cost drivers. Future research will focus on evaluating hybrid TabNet-Boosting architectures and integrating longitudinal temporal claims to capture evolving risk profiles across multinational cohorts.

Hybrid Measured-Synthetic Framework for Daily Energy Prediction of Rooftop Solar PV Systems in Tropical Regions

Nur Iksan — 2026-09-15

Rooftop photovoltaic (PV) systems are increasingly deployed in tropical regions to support renewable energy adoption. However, reliable PV performance assessment typically requires long-term field measurements, which are often unavailable due to limited monitoring infrastructure and operational constraints. This study proposes a hybrid measured-synthetic framework for generating representative PV performance profiles and predicting energy production using limited field observations combined with meteorological data. The main contribution of this work is the development of a practical framework that extends sparse PV measurements into daily, monthly, and annual performance estimates. The proposed framework integrates measured PV operating parameters, including module temperature, voltage, current, and wind speed, collected from a rooftop PV installation in Batam, Indonesia, with solar radiation and ambient temperature data obtained from the NASA POWER database. Due to the limited availability of field measurements, a synthetic hourly dataset was generated using actual observations collected on 15 October 2022 in conjunction with photovoltaic performance modeling and meteorological information. The generated dataset was subsequently used to reconstruct representative daily operating profiles and estimate energy production over longer time horizons. The results indicate that the proposed framework successfully generated physically consistent PV performance profiles under tropical climatic conditions. The framework estimated a daily energy production of 34.27 kWh/day and an annual energy production of approximately 12,505.5 kWh/year. Validation against the available field measurements produced Mean Absolute Percentage Error (MAPE) values of 0.08%, 6.80%, and 7.03% for voltage, current, and power, respectively, indicating satisfactory agreement between measured and synthetic datasets. The proposed framework reduces dependence on extensive monitoring campaigns while preserving representative operational characteristics of photovoltaic systems. Therefore, it can serve as an effective tool for preliminary PV performance assessment and energy forecasting in regions where long-term measurement data are unavailable. Future work should incorporate longer monitoring periods, direct irradiance measurements, and broader validation datasets to improve prediction accuracy and generalizability.

Predicting Customer Purchase Intent in E-Commerce Platforms Using Session-Based Logistic Regression Analytics

Nor Anisah Salamah Mohd Rosli — 2026-09-15

The rapid acceleration of global e-commerce has fundamentally transformed retail operations, yet digital merchants continue to confront substantial financial losses resulting from inefficient inventory planning, cart abandonment, and imprecise consumer behavior forecasting. This research develops ShopPredict, a session-based predictive analytics framework designed to forecast online shopping purchase completion by modeling user interaction dynamics including product viewing frequency, cart additions, session price averages, and unique item explorations using a Logistic Regression classification approach. Utilizing an empirical multi-category e-commerce dataset comprising millions of user session logs, the classification model was trained and evaluated across structured behavioral features. The experimental results demonstrated a high overall accuracy of 93.48%; however, rigorous metric decomposition revealed a critical accuracy paradox, wherein the model achieved a high precision of 0.94 for non-purchase sessions but exhibited a low recall of 0.16 and an F1-score of 0.25 for purchase events due to severe underlying class imbalance. System usability and functional architecture evaluations yielded positive user acceptance scores averaging above 4.0 on a 5-point Likert scale. This study underscores the utility of session-based linear probability modeling while demonstrating that high nominal accuracy in e-commerce prediction can be misleading without rigorous class-rebalancing strategies. Future research must incorporate advanced ensemble architectures, SMOTE oversampling, and dynamic session features to optimize purchase intent detection in real-time retail environments