Projects

This project involved conducting a detailed social media network analysis using R on posts tagged with #WSUCOMP7025. The analysis focused on identifying key user interactions, network structure, and influence patterns within the student community on Mastodon. Techniques such as graph creation, network decomposition, and PageRank analysis were applied to determine central figures and the connectivity within the network. Additionally, expected payoff calculations were performed to assess strategic account selection, providing insights into the distribution and influence of students across different nodes in the network.

Github: github.com/rizwankarimrafat/Social-Media-Network-Analysis-Using-R-for-WSUCOMP7025-Hashtag

This project explores and compares the performance of Conditional Variational Autoencoders (C-VAE) and Conditional Generative Adversarial Networks (C-GAN) in generating Japanese characters from the Kuzushiji-49 dataset. The models were evaluated on their ability to produce images conditioned on character class and style. The analysis involved both qualitative (visual) and quantitative assessments, including metrics such as Mean Squared Error (MSE) and Structural Similarity Index (SSI). The project provides insights into the strengths and limitations of each model in capturing and replicating fine details in character generation tasks.

Github: github.com/rizwankarimrafat/Comparative-Analysis-of-C-VAE-and-C-GAN-for-Japanese-Character-Generation

This project involved analyzing a dataset of NSW government schools to predict outcomes such as "Opportunity Class" and "Late Opening School" using machine learning techniques. The models implemented include Neural Networks, Support Vector Machines (SVM), and Naive Bayes. The project focused on data preprocessing, model training, and evaluation to determine the most accurate method for predicting school-related outcomes. Each model's performance was compared based on accuracy, precision, recall, and F1-score, with insights drawn from the results to identify the best-performing approach.

Github: github.com/rizwankarimrafat/Predictive-Modeling-of-NSW-School-Data-Using-Neural-Networks-SVM-and-Naive-Baye

This project involved the development and evaluation of various predictive models to forecast daily temperature using linear regression, gradient descent, logistic regression, and ridge regression techniques. The analysis included binary and multiclass classification tasks, where temperature data was classified based on different thresholds. The models were compared based on metrics like R², mean squared error (MSE), and accuracy, with a focus on improving model performance through data preprocessing and parameter tuning.

Github: github.com/rizwankarimrafat/Predictive-Modeling-and-Classification-of-Temperature-Data

This project involved applying advanced statistical methods to analyze a dataset of cat births. Key tasks included modeling the density of cat births throughout the year, examining the relationship between cat breeds and family income, analyzing the correlation between gestation periods and income, and identifying birth location trends. The project utilized techniques such as density estimation, regression mixture modeling, and multivariate normal mixture models to derive insights from the data.

Github: github.com/rizwankarimrafat/Advanced-Statistical-Analysis-of-Cat-Demographics-and-Birth-Trends

This project involved analyzing a ride-sharing dataset to explore various statistical relationships. The analysis included hypothesis testing, confidence interval estimation, and linear regression to investigate associations between variables like passenger ratings, driver tips, and fare amounts across different days and conditions. Key statistical techniques such as chi-squared tests, bootstrapping, and ANOVA were applied to derive insights from the data.

Github: github.com/rizwankarimrafat/Statistical-Analysis-of-Ride-Sharing-Data

Conducted a comprehensive study on enhancing traditional CCTV systems with AI-driven gait recognition technology. The project focused on utilizing existing camera infrastructure to identify individuals based on their walking patterns, combining motion analysis with advanced machine learning models for improved security surveillance.

Developed an IoT-based system for automating notice boards in public spaces. The project used Arduino, GSM modules, and mobile apps for remote message updates and SMS notifications. Conducted a literature review to evaluate the effectiveness of IoT in enhancing traditional notice boards, focusing on eco-friendly and cost-effective solutions.