Project Report (Bachelor of Science in Computer Science and Engineering)

Permanent URI for this collectionhttps://hdl.handle.net/10361/28441

Browse

Now showing 1 - 4 of 4

Analyzing income inequality in South Asia through interactive data visualizations
(BRAC University, 2026-01) Oishe, Shaiera Sultana; Mayami, Asma Faria; Mafij, Tabassum Bintay; Mehzabeen, Samin; Rasel, Annajiat Alim; Rakin Bin Rabbani; Department of Computer Science and Engineering
In South Asia, where socioeconomic gaps have frequently been widened rather than addressed by fast economic expansion, income inequality is still a major problem. Static charts and technical reports are two examples of traditional means of presenting inequality statistics, and they often fall short in explaining the wide-ranging, demographic, and region-specific subtleties of inequality to non-experts and local officials. By creating, refining, and testing an interactive data visualization tool that converts intricate inequality indicators into understandable, captivating, and useful visual narratives, our study fills this communication gap. This study incorporates indices of economic, educational, gender, and digital inclusion across South Asian nations, using data from the World Bank, UNDP, national government websites, and gender-disaggregated databases. Utilizing visualization technologies like Tableau and Streamlit and Python libraries like Pandas, Plotly, and GeoPandas, the project delivers an interactive dashboard with real-time, multi-dimensional filtering and comparative analysis capabilities of wide-ranging datasets. Choropleth maps, time series graphs, and dynamic bar charts are some of the visualizations used to display both temporal and spatial shifts in inequality. According to user testing with journalists, students and development professionals, the benefits of interactive visual storytelling greatly enhance the understanding, engagement and memory of inequality statistics, especially for non-technical and novice audiences. Additionally, data literacy and public discussion is enabled by design features such as mounting narrative on maps, geographical filtering and simplicity. This study comes to the conclusion that data visualization is a potent instrument for altering the method with which the public interacts with social and economic facts. Most importantly, this gives a repeatable paradigm for data-driven policy support, public awareness, and inclusive governance in South Asia.
SafeConnect: an inclusive personal safety, connectivity and real-time emergency alert app
(BRAC University, 2025-02) Arnob, Rayean Mahmud; Alam, Md. Ashraful; Department of Computer Science and Engineering
"Developing effective personal safety tools has become crucial in response to rising safety concerns in today’s world. This project focuses on the development and evaluation of SafeConnect, a mobile application designed to enhance personal safety, improve connectivity, and provide rapid emergency response in real-time. SafeConnect’s goal is to equip individuals with tools for proactive risk management, which helps to access fast support in urgent situations. This app is developed using Flutter framework which provides seamless cross-platform experience. The core functionality of this app includes key features such as emergency contact integration, real-time chat with guardians, and location-based services through interactive maps. Users can trigger a high-priority emergency alert message with their real-time location de-tails, by pressing a panic button or shaking their phone. Additional features include anonymous incident reporting, access to inspirational content and safety guidelines. In addition, SafeConnect uses Firebase to connect in real-time to the database, as well as the Google Maps API for obtaining accurate geolocation. SafeConnect connects emergency requirements and technological solutions together to fasten the process for people to respond to an emergency."
Sales forecasting using machine learning
(BRAC University, 2024-10) Nabil, Sadman Sakib; Islam, Md Tanvir; Muhit, Sadman Aziz; Chakrabarty, Amitabha; Department of Computer Science and Engineering
In today’s aggressive and fast-paced economy, the ability to forecast sales accurately and effectively denotes a proper utilization of the available resources in planning. Typical sales forecasting methods fail quite often to measure the dynamic market environment owing to the fact that they are totally influenced by past data and also expert opinion. Therefore, this research seeks to validation of sales forecast accuracy with respect to the integration of machine learning (ML) in enhancing its capability. Considering available historical sales figures and some social media trends, machine learning techniques are able to provide realistic and satisfactory forecasts. The paper discusses the advantages of machine learning (ML) to the old methods, for instance, quick detection of the emerging trends, dealing with big data, and adaptation to the situation. Some problems, such as data quality and system integration are also considered. Some of these include ensemble methods, neural networks, and regression, and such techniques are used in machine learning. This article discusses how the integration of machine learning (ML) in sales forecasting will help companies in management and decision making leading to better performance compared to competitors.
Loan approval prediction using machine learning algorithms
(BRAC University, 2024-10) Roy, Reak; Alam, Tahsin; Kabir, Syed Hafiz; Awsaf, Mirza Abyaz; Haque, Shadik Ul; Chakrabarty, Amitabha; Department of Computer Science and Engineering
This research describes the potential of several classifiers of classical machine learning and architecture of deep neural networks when predicting the status of a loan application. The data set of 613 observations and 13 features, provided with the information about the applicants and their credit profiles, was utilized together with other techniques, such as bootstrapping, for more data qualityutimaltely leading to 9824 observations. Some imputation strategies were applied to deal with the lack of values, while also features were carefully prepared by employing ANOVA, Mutual Information and Tree based approaches among other statistical methods. For the validation of the model performance, the dataset was split into two parts: training (70%) and testing (30%). Many classical machine learning algorithms were applied including but not limited to Logistic Regression, Support Vector Classifiers(SVC), Decision Trees, Random Forests, Multi-Layer Perceptron, Gradient Boosting machines, K-Nearest Neighbors, etc. Out of all models used in the research, Random Forest Classifier demonstrated the most high values of accuracy of 86.84% and F1- score (0.9043), hence it was the best performing one. Advanced methodologies such as SMOTE (accuracy of 88.16%) and ADASYN (accuracy of 87.07% )were also used to handle the issue of class imbalance, where the performance of K- Nearest Neighbors was impressive acuuracy of 88.16% after resampling. In a different, yet similar analysis, five types of neural network architectures, Simple Recurrent Neural Network(RNN), Long-Short Term Memory(LSTM), Convolutional Neural Networks( CNN), Fully Connvolutional Neural Networks(FCNN) and Fully Connected Neural Networks(FCN) were built with the use of Tensorflow, Scikit-learn, and Numpy running on Google Colaboratory notebooks. The outcomes showed that the Fully Convolutional Network (FCN) has the best validation accuracy of 89.75% and validation loss of 0.2255 among the models built.

Browse

Recent Submissions