Projects
These are some of my coolest projects. Check it out!
Machismo: Online Misogyny in Philippines through the years
Bayaborda, Punzalan, Ramos, Timajo
2022
This is a thesis project done at the Asian institute of Management. We were tasked by one of the institute's clients (Rappler) to investigate the nature of misogyny in Philippine society using Natural Language Processing (NLP) Techniques. We utilized Gibbs Sampling Dirichlet Multinomial Mixture (GSDMM) and other unsupervised learning models for NLP Problems such as Text Clustering. We also applied the optimized Support Vector Machine (SVM) model for multi-lingual sentiments analysis and performed Statistical Analysis to solve Big Data Problems for single node machine setups and lower-power workstations
Where Is the Love? Identifying Hate Speech in Philippine
Election-Related Tweets
Baluyut, dela Resma, Fernando, Lapid, Sy, Timajo
2022
The study utilized a dataset of tweets scraped during the 2016 Philippine electoral campaign that are labeled as hate and non-hate speech. Data was cleaned and processed using various NLP methods such as TF-IDF, then various machine learning classifier models were built including linear, tree-based, and multilayer perceptron models. Hyperparameter tuning was performed to arrive at the best models based on accuracy and runtime. The best model was then used to test another dataset made up of recently scraped and manually labeled tweets about the 2022 elections. Finally, SHAP was employed to understand the factors that contribute to hate speech on a global scale, and another interpretability method, LIME, was used to interpret the results on a local level.
Complex Portfolio Building: A network analysis of PSEi Composite Index Stocks
Bayaborda, Punzalan, Ramos, Timajo
2022
This project is a Network Science and Graph Theory based project on the Philippine Stock Exchange Composite Index. We performed Stock Correlation Network building and subsequently performed Network Analysis on the formed graph. There are tons of insights uncovered from this project. We used data from the Philippine Stock Exchange to build a network that uses organizations as nodes and their corresponding stock price-based Pearson-Correlation as the network edge to determine the relationship between the stocks. This allowed them to uncover
network properties that lead to a better understanding of how the stock market moves. This meant that better decisions when building a stock portfolio.
Neuroscan.io: An MRI Scan Interpretation device for Brain Tumors
Austria, Bayaborda, Punzalan, Timajo
2022
This study is an attempt to make Radiologists and Neurologists tasks much more efficient by automating the interpretation of MRI Scans. The scanner can detect whether or not a patient's MRI Scan contains a tumor and if there is a tumor further detect what kind of tumor is present on the MRI Scan. This approach is an attempt to minimize interpretation overhead and at the same time use Deep learning class precision to detect different types of tumor if they are present at all. The best performing model is extracted from a test of different CNN architectures and is deployed on a Web Application with the help of the Flask backend framework with a simple HTML user interface. The project was also deployed using Flask and AWS during the pre and post public presentation.
.jpg)
Financial Health Metrics: Predicting Company Bankruptcy
Angelo S. Timajo, MSc
2021
This project shows the ability of machine learning to predict potential company bankruptcy using multiple features. The Taiwanese Companies Bankruptcy Dataset was sourced from the UCI ML Repository. Advanced techniques were applied here such as PCA and even SMOTE to solve some problems with the dataset such as high dimensionality and imbalanced nature. Most features contained in this data are financial management indicators which follow the Accounting formula Assets = Liabilities + Equity, therefore, the data when checked, is linearly correlated to each other.
