Umm e Hani's Projects
This project implements a neural network architecture using separate processes and threads on a multi-core processor. It utilizes inter-process communication through pipes for exchanging weights and biases between processes. Implemented in C++, it leverages multi-core processing for efficient training. Designed for Linux environments.
The project involves creating a web scraper to collect detailed information about products in the Amazon 'Fashion' category. The goal is to gather data on at least 1000 products from each subcategory (e.g., shoes, clothes, watches) and organize it in JSON format for trend analysis and insights into popular items.
This project creates a live product recommender system using the Amazon dataset. It involves data processing with Apache Spark, MongoDB, and exploratory analysis. A recommendation model (ALS) is trained and integrated into a Flask web application, with real-time updates via Apache Kafka.
This project explores the daily price fluctuations of Bitcoin using time series analysis. It compares the performance of linear and non-linear regression models, as well as ARIMA, to predict future prices. The goal is to determine the most suitable model for forecasting Bitcoin prices.
This project develops a forecasting system using ARIMA, ANN, and hybrid models for finance, energy, and environmental sectors, featuring a user-friendly interface for data visualization and analysis.
Data Warehouse (DW) prototype for Electronica Business, to analyze customer shopping behavior in near-real-time. It uses the HYBRIDJOIN algorithm in a multithreaded Java implementation to efficiently join and transform transactional data for the DW. The project aims to optimize sales strategies by providing insights through various OLAP queries.
Performing Exploratory Data Analysis (EDA) on the Berkeley Earth dataset plus two more time series datasets , focusing on a specific region to analyze climate trends. The project includes creating Tableau dashboards and stories to visually represent the findings, emphasizing beautiful design and insightful storytelling.
This project explores the Cornell Movie Dialogs Dataset through NLP and interactive visualizations. Using Python and libraries like NLTK for data processing, and D3.js for web-based visualizations, it presents sophisticated, interactive graphs and tree visualizations. The project integrates Flask for seamless data transfer.
A comprehensive C++ project implementing image processing operations such as statistics computation, connected component extraction, run-length coding, quad tree representation, and shape recognition. Utilizing data structures like arrays, queues, stacks, and linked lists.
This project involves implementing interactive foreground segmentation using K-Means clustering and face recognition using a k-NN classifier. It includes tasks such as image segmentation with Lazy Snapping, and face recognition on the CMU PIE dataset using k-NN, SVM, and GaussianNB classifiers.
Creating an intelligent dashboard for graph layout using d3.js. It allows users to upload an adjacency matrix, analyze the graph's characteristics, and classify it as a General Graph, DAG, or Tree. Based on the graph type, the dashboard provides appropriate layout options for visualization, implemented with custom layout algorithms from scratch.
This project implements a real-time data processing pipeline using Apache Kafka, Python, and MongoDB. It processes Microsoft and Apple stock data, with Kafka consumers analyzing price differences, risk, and percentage changes, then storing results in MongoDB.
Implementation of Locality-Sensitive Hashing (LSH) with Mel-Frequency Cepstral Coefficients (MFCC) features for audio duplicate detection and explores its application using a Flask web application.
Web application for meme sentiment analysis using machine learning classifiers (text & image) with Flask deployment.
This is an implementation of a multiplayer game where players navigate a two-dimensional grid to collect items. Players can move concurrently on the game board and collect items to earn points. The game is implemented using threads to handle player movement and item collection, ensuring a seamless and engaging gameplay experience.
This project implements a scalable algorithm to rank nodes in a network based on their proximity to user-specified source nodes. It leverages Hadoop's distributed processing power to handle large networks efficiently. The core algorithm, implemented in Python using MapReduce, personalizes traditional PageRank.
The repository provides solutions to two concurrency challenges in C/C++. The first involves implementing a lock and unlock mechanism using the Bakery algorithm for mutual exclusion. The second addresses traffic management on a bridge, ensuring limited capacity is maintained using mutex synchronization.
This repository contains solutions to operating system assignments covering process management, inter-process communication, and system administration tasks using shell scripting. Explore for practical implementations and explanations.
PWB is a versatile project management and issue tracking tool tailored for agile development teams. It offers comprehensive features including customizable workflows, detailed reporting, and user-friendly collaboration tools. PWB enables efficient project setup, task management, and progress tracking, ensuring high-quality software delivery.
This project is a retail billing system utilizing YOLOv8, trained on a custom dataset of clothing, accessories, and supermarket items. The dataset was personally curated, featuring photos from personal collections and supermarket visits. The trained YOLOv8 model detects objects in images to generate accurate bills efficiently.
sentriX is a web application that leverages Flask and React to perform sentiment analysis on text. Utilizing the TextBlob library, the application analyzes user-input text to determine whether it is positive, negative, or neutral. The application features four main pages: Home, Tool, About, and Contact.
Config files for my GitHub profile.