Recent Advances in Computer Science and Communications - Volume 18, Issue 7, 2025

In today’s fast-paced world, online shopping has become a staple activity, especially for busy individuals seeking convenience. Online shopping has emerged as a predominant activity, particularly popular among the younger demographic. The hectic lifestyles of the working class have elevated online shopping to a convenient and often essential practice, providing a respite amidst their busy schedules. Simultaneously, businesses in the competitive E-commerce market recognize the importance of understanding customer behavior and perceptions to foster loyalty and ensure sustainability.

In response to the evolving landscape of online shopping, we have undertaken a comprehensive analysis using machine learning techniques. Our approach involves the utilization of machine learning algorithms to recognize patterns and make precise predictions. We divide the data set into quarters, assess sales income per quarter, and further partition the data into training and testing sets. The subsequent steps involve forecasting revenue for upcoming quarters and identifying top-performing commodities to devise a Python-based model for strategic customer retention. The methodology begins by segregating the sales data set into quarters, followed by the calculation of quarterly sales income. Using a machine learning system, our approach forecasts revenue for future quarters and identifies high-performing commodities based on quarterly sales rates.

The culmination of these results leads to the development of a Python-based model for strategic customer retention. The outcome of our analysis not only facilitates precise predictions of future revenue but also contributes to the creation of a strategic model for customer retention.

By identifying high-performing commodities and leveraging machine learning algorithms, we establish a symbiotic loop of sales and purchasing between customers and the E-commerce company. This algorithmically-driven loop promises incremental profitability for both the customers and the E-commerce company, creating a symbiotic relationship.

The use of 20% blended biofuels to fossil fuels is one of the important targets of the Government of India to address the impacts of Climate Change, energy-related environmental pollution, and illnesses due to air pollution.

The National Policy on Biofuels 2018 (NPB 2018) is in place to boost the emerging production of biofuels and, therefore, respond to different international agreements, including the Sustainable Development Goals (SDGs) and the Paris Agreement. Hence, this article examined the production of biofuels in India in line with Agenda 2030 to project the share to be taken by biofuels as its contribution to the country’s energy needs.

The results were compromising; it was observed that the data from 2000 up to 2017 were not on the side of realizing the targets of production and consumption of biofuels in India, whereas the data from 2018 up to now showed a hope of achieving 2030 set goal of E20 petrol in 2025-26, and E5 diesel in 2030. It was clear that the production of bioethanol was boosting compared to its sibling biodiesel, and renewable energy will continue to have a hard take a good share in the total annual energy used in India.

It is recommended to share data between different stakeholders to promote more research, as the low performance in achieving the targets was due to poor communication and missing technology, rather than the lack of feedstock or unavailability of production facilities.

Cyber-attacks related to ransomware have increased significantly in cloud manufacturing industries over the last decade, causing considerable disruptions to organizations. This type of information may also include personal details, patent rights, bank account details, etc. This type of malware requires new and better mitigation methods.

The primary objective of this study was to provide an algorithm to execute experiments using genetic algorithms for load balancing with decision trees and mitigate ransomware.

Hybrid analysis and machine learning techniques were used in this study to identify ransomware. Since a wide range of samples impacted by ransomware share most of the characteristics, it may be possible to use this study to detect current and future malware variants in industries.

In patented industrial technology, ransomware mitigation plays a crucial role based on the analysis of various papers and patents-to-science references. In this study, a machine learning mitigation algorithm, GeniLeaf Decision Tree (GLDT), was applied to a featured dataset using a genetic algorithm with a decision tree.

Machine learning and load balancing are used to gain insight into ransomware behavior. By using the proposed approach to mitigate ransomware and spoofing patterns, a high level of accuracy is achieved. Using GeniLeaf Decision Trees to mitigate ransomware is a significant innovation.

To address the challenges of low visibility, object recognition difficulties, and low detection accuracy in foggy weather, this paper introduces the WVIT-YOLO real-time fog detection model, built on the YOLOv5 framework. The NVIT-Net backbone network, incorporating NTB and NCB modules, enhances the model's ability to extract both global and local features from images.

An efficient convolutional C3_DSConv module is designed and integrated with channel attention mechanisms and ShuffleAttention at each upsampling stage, improving the model's computational speed and its ability to detect small and blurry objects. The Wise-IOU loss function is utilized during the prediction stage to enhance the model's convergence efficiency.

Experimental results on the publicly available RTTS dataset for vehicle detection in foggy conditions demonstrate that the WVIT-YOLO model achieves a 3.2% increase in precision, a 9.5% rise in recall, and an 8.6% improvement in mAP50 compared to the baseline model. Furthermore, WVIT-YOLO shows a 9.5% and 8.6% mAP50 improvement over YOLOv7 and YOLOv8, respectively. For detecting small and blurry objects in foggy conditions, the model demonstrates approximately a 5% improvement over the benchmark, significantly enhancing the detection network's generalization ability under foggy conditions.

This advancement is crucial for improving vehicle safety in such weather. The code is available at https://github.com/QinghuaZhang1/mode.

Cryptocurrency is the new mode of transaction that is slowly replacing the conventional assets of transactions. With the rise and advancement in the technology called blockchain and the volatile nature of the market, cryptocurrencies are showing possibilities of making huge profits. This new trend is slowly becoming accessible to all classes of society in every part of the world while creating a great opportunity for scholars and analysts to conduct new research. However, unlike the stock market, the prices of cryptocurrencies show highly volatile and dynamic behaviour.

Several research in the past have been performed to predict the nature and other aspects of cryptocurrencies with the aid of machine learning to forecast the future market based on past data. In the proposed work, 6 different machine-learning models have been used for forecasting the prices of cryptocurrencies. The algorithms used are Autoregressive Integrated Moving Average (ARIMA), Gated Regression Unit (GRU), Long Short-Term Memory (LSTM), Bi-LSTM and Vector Autoregression (VAR).

Further, the accuracies of different models are compared and analysed. Furthermore, these models are used to predict the first-order difference between 24 hours of all-time high and low values. This evaluation gives us an idea of where the price band of the currency and where is it most likely to be present during any 24-hour timespan.

This can effectively lead to better decision making in choosing crypto-currency. The GRU model achieved the best performance in cryptocurrency prediction with an RMSE of 0.2242 and MAE of 0.1598 for BTC prices, while the LSTM followed closely with an RMSE of 0.3473 and MAE of 0.2557. Both models outperformed ARIMA, Bi-LSTM, VAR, and FB Prophet, which showed significantly higher error rates.