Water quality is a critical factor in shrimp farming, and the success of shrimp production is closely tied to the overall condition of the water. Challenges such as rapid population growth, environmental pollution, and global warming have led to a decline in fisheries production, particularly in the freshwater shrimp sector. This study addresses these challenges by monitoring multiple water parameters in shrimp farms, including pH, temperature, TDS, EC, and salinity. Traditional manual monitoring systems are known to be cumbersome, time-consuming, and lacking real-time capabilities. Consequently, a continuous and automated monitoring system becomes imperative for efficient and real-time metrics handling. This study introduces a real-time freshwater shrimp (locally named Galda, i.e., Macrobrachium Rosenbergii) farm monitoring system. The proposed system incorporates technologies such as microcontroller-based physical devices, IoT, cloud storage with service, machine learning models, and web applications. This integrated system enables users to remotely monitor shrimp farms and receive alerts when water parameters fall outside the optimal range. The physical implementation involves a set of sensors for collecting data on water metrics in shrimp farms. Regression analysis is employed for predicting next-day values, and a newly developed decision-based algorithm classifies shrimp production levels into low, medium, and maximum categories using six well-known classification algorithms. The system demonstrates a high success rate for next-day predictions (r² of 0.94) by multiple linear regression, and the accuracy in classifying shrimp production is 97.84 % by Random Forest. Additionally, a 'Smart Aquaculture Analytics' web application has been developed, offering features such as real-time dashboards, historical data visualization, prediction and classification tools, and automated notifications to farmers in Bangladesh.

Food delivery methods are at the top of the list in today's world. People's attitudes toward food delivery systems are usually influenced by food quality and delivery time. We did a sentiment analysis of consumer comments on the Facebook pages of Food Panda, HungryNaki, Pathao Food, and Shohoz Food, and data was acquired from these four sites’ remarks. In natural language processing (NLP) task, before the model was implemented, we went through a rigorous data pre-processing process that included stages like adding contractions, removing stop words, tokenizing, and more. Four supervised classification techniques are used: extreme gradient boosting (XGB), random forest classifier (RFC), decision tree classifier (DTC), and multi nominal Naive Bayes (MNB). Three deep learning (DL) models are used: convolutional neural network (CNN), long term short memory (LSTM), and recurrent neural network (RNN). The XGB model exceeds all four machine learning (ML) algorithms with an accuracy of 89.64%. LSTM has the highest accuracy rate of the three DL algorithms, with an accuracy of 91.07%. Among ML and DL models, LSTM DL takes the lead to predict the sentiment.

Cryptocurrencies are encrypted digital or virtual money used to avoid counterfeiting and double spending. The scope of this study is to evaluate cryptocurrencies and forecast their price in the context of the currency rate trends. A public survey was conducted to determine which cryptocurrency is the most well-known among Bangladeshi people. According to the survey respondents, Bitcoin is the most famous cryptocurrency among the eight digital currencies. After that, we'll explore the four most well-known cryptocurrencies: Bitcoin, Ethereum, Litecoin, and Tether token. The 'YFinance' python package collects our cryptocurrency dataset, and the relative strength index (RSI) is employed to investigate these cryptocurrencies. Autoregressive (AR), moving average (MA), and autoregressive moving average (ARMA) models are applied to our time-series data from 2015-1-1 to 2021-6-1. Using the 'closing' price and a simple moving average (SMA) graph, bitcoin and tether are identified as oversold or overbought cryptocurrencies. We employ the seasonal decomposed technique into the dataset before implementing the model, and the augmented dickey-fuller test (ADF) indicates too much seasonality in the dataset. The autoregressive (AR) model is the most accurate in predicting the price of Bitcoin, Ethereum, Litecoin, and Tether-token, with 97.21%, 96.04%, 95.8%, and 99.91% accuracy, consecutively.

Bengali, an Indo-Aryan language, features a complex grammatical structure with tenses, which is crucial for natural language processing (NLP) applications like text classification, machine translation, and sentiment analysis. The BanglaTense dataset is a large-scale, meticulously curated collection of Bangla sentences categorized by their tense: Past, present, and future. Addressing the resource gap in NLP for the Bangla language, BanglaTense provides a curated resource for Bangla sentence classification, featuring 17,819 annotated sentences, with 5,629 in the past tense, 6,101 in the present tense, and 6,089 in the future tense. This dataset is a benchmark for evaluating NLP models on Bangla sentence classification, promoting linguistic diversity and inclusive language models while ensuring balanced representation across categories. Preprocessing steps are applied to enhance data quality, including anonymization and duplicate removal. Three native Bangla speakers independently assessed the tense labels of the sentences, ensuring the dataset's reliability. BanglaTense is designed to advance research and development in NLP for Bangla, offering valuable applications in tense detection, text classification, language modeling, and educational tools. This dataset supports linguistic study and enhances the development of precise and context-aware NLP models by providing a robust foundation for temporal analysis in Bangla sentences. The dataset is openly available for academic and research purposes, promoting collaboration and innovation within the Bangla NLP community.

Chronic kidney disease (CKD) is characterized by persistent abnormalities in urinary biomarkers or reduced renal function, posing risks not only of progression to end-stage kidney disease but also of accelerated cardiovascular complications and mortality. The use of computer-aided automated diagnostics can assist nephrologists in early detection and accurate classification, which are essential for improving patient outcomes. This study utilized clinical features of CKD to develop and evaluate six base machine learning classifiers (logistic regression, K-nearest neighbors, AdaBoost, decision tree classifier, random forest, and multilayer perceptron) alongside two novel ensemble models (MKR Stacking and MKR Voting) for CKD prediction and classification. The proposed models were trained on five pre-processed CKD datasets using four robust feature selection techniques, including Lasso, Fisher score, Information Gain, and Relief. The models’ performance was assessed using accuracy, precision, recall, F1-Score, error rate, AUC, and computational time. Among the tested algorithms, MKR Stacking achieved the highest accuracy of 99.50%, outperforming Random Forest (98.75%) and MKR Voting (98%). The XAI technique SHAP and model validation on another CKD dataset highlight the superior prediction capabilities of the proposed ensemble methods compared to traditional classification algorithms. The study further advocates for integrating high-performing models into the Internet of Medical Things and Robotic Process Automation frameworks, enabling real-time monitoring, predictive analytics, and efficient CKD diagnosis. Such integration has the potential to transform CKD management, facilitating early interventions and personalized treatment plans through advanced machine-learning applications.

Abnormal brain tissue or cell growth is known as a brain tumor. One of the body's most intricate organs is the brain, where billions of cells work together. As a head tumor grows, the brain suffers damage due to its increasingly dense core. Magnetic resonance imaging, or MRI, is a type of medical imaging that enables radiologists to view the inside of body structures without the need for surgery. The image-based medical diagnosis expert system is crucial for a brain tumor patient. In this study, we combined two Magnetic Resonance Imaging (MRI)-based image datasets from Figshare and Kaggle to identify brain tumor MRI using a variety of convolutional neural network designs. To achieve competitive performance, we employ several data preprocessing techniques, such as resizing and enhancing contrast. The image augmentation techniques (E.g., rotated, width shifted, height shifted, shear shifted, and horizontally flipped) are used to increase data size, and five pre-trained models employed, including VGG-16, VGG-19, ResNet-50, Xception, and Inception-V3. The model with the highest accuracy, ResNet-50, performs at 96.76 percent. The model with the highest precision overall is Inception V3, with a precision score of 98.83 percent. ResNet-50 performs at 96.96% for F1-Score. The prominent accuracy of the implemented model, i.e., ResNet-50, compared with several earlier studies to validate the consequence of this introspection. The outcome of this study can be used in the medical diagnosis of brain tumors with an MRI-based expert system.