Projects
Application of Natural Language Processing for Phishing Detection Using Machine and Deep Learning Models
The prevalence of phishing scams on the Internet served as the inspiration of the research article. The study utilized Machine Learning techniques such as Logistic Regression, Multinomial Naive Bayes, and Deep Learning models such as Long Short Term Memory (LSTM), Gated Recurrent Units, and their bidirectional counterparts on the procured dataset. The results were satisfactory but the LSTM and GRU had the highest training and validation score with an overall accuracy of 95%.
[Publication Site]
Sentiment Analysis on Ice Skating Reviews in Philippines
This is a notebook project conducted to determine if I can expect to have a good time ice skating in the aforementioned country which is likely based on the results. For this project, review data from different branches (i.e., Mall of Asia, Megamall, Seaside City Cebu) were scraped from Google Maps which were then cleaned. Subsequently, the cleaned data were used to perform exploratory data analysis and sentiment analysis with visualizations.
[GitHub Repo]
Design of Footage Filtering System for CCTV Archives using Object Detection
This project was made for the completion of our capstone for Design Project course in our program. The general objective of this project is to design a CCTV footage filtering system through human clothing and weapons using object detection. Furthermore, the system aims to assist in finding the relevant clips from the footage. PyQt6 was used for implementing the algorithm and YOLOv5 from Ultralytics for the detection algorithm. Moreover, it uses fast-colorthief for classifying the color of detected clothings.
[GitHub Repo]
IoT Data Pipeline
This project implements an Extract-Transform-Load (ETL) pipeline for IoT data using Kafka for streaming and MySQL for storage. It is designed to handle real-time ingestion of IoT device data, process it, and persist it into a relational database for further analytics. The project uses Docker to run Kafka and MySQL. Although the mode of data fetching is via API, the project is designed to easily change the method of extraction, possibly directly from a machine PLC. Thus, this project is a good template for persisting real-time IoT data.
IoT Data Source: https://thingspeak.mathworks.com/channels/1293177/feeds
[GitHub Repo]
CDC for Job Order Tracking System
A real-time monitoring system that tracks manufacturing job orders by capturing live database changes using Debezium CDC and streaming them through Apache Kafka. The system processes the event streams and visualizes updates instantly on a Streamlit dashboard, providing real-time visibility of production progress without manual refresh.
[GitHub Repo]
Prediction of House Prices in the Philippines
This project aims to predict house prices in the Philippines given the following factors: Number of Bedrooms, Number of Bathrooms, Floor Area, Land Area, and Urbanicity. For data wrangling and feature engineering, Numpy and Pandas were used. For Machine Learning modeling, Scikit-Learn was used wherein hyperparameter optimization across three models were performed. These models are RandomForestRegressor(), SVR(), and StackingRegressor(). It was observed that the errors were lower for houses with lower prices which was reflected by the target values because its distribution is positively skewed, that is - contains more houses with lower prices. A basic Streamlit application was developed to demonstrate the deployment of the model.
[App Demo] [GitHub Repo]
Time Series Analysis and Forecasting on Coca-cola Stocks
This project aims to explore time series analysis on stocks data. It also aims to explore time series forecasting using non-deep learning (i.e., ARIMA and SARIMAX) and deep learning (i.e., Long Short Term Memory (LSTM)) techniques. Various error metrics were used to asses the performance of each technique wherein LSTM had the least value across all error metrics. Thus, it was used to perform the final step which is forecasting.
[GitHub Repo]
Create, Read, Update, and Delete (CRUD) App - CSIBALE
This project was made for the completion of my Software Design course along with my colleagues. The client is a non-government organization (NGO) with a mission to assist and create opportunities for children with disabilities which are their beneficiaries. To conveniently keep records of their beneficiaries' information, this Create-Read-Update-Delete (CRUD) application was developed.
[GitHub Repo]
SpotInsights
I am interested on having a look on my music taste and I happen to stumble upon an application called "Receiptify" which exactly does that. Thus, I made an application inspired from it. This is a Flask application that provides insights and visualizations regarding the user's tracks and artists played on Spotify using their API.
[GitHub Repo]
Coniculator
This application was my final project during my second-year college for my Object Oriented Programming (OOP) course. It is a PyQt5 Application for calculating values (e.g., center, foci, vertices, length of latus rectum, eccentricity, etc.) needed given a template conic equations to plot the following conic shapes: Circle, Ellipse, Parabola, Hyperbola.
[GitHub Repo]
