Denyse

• Researched different language detection models and devised a framework with FastText to differentiate addresses containing Thai and English characters for various use cases, achieving a 99.8% accuracy
• Built data pipelines using PySpark to process big data containing Thailand addresses from multiple sources to a target database - incorporated language detection, hierarchical clustering, NER model, and reverse geocoding of addresses
• Explored Shopee’s Thailand polygon dataset and utilised Kepler.gl to verify if admin division levels can cover the entire country

• Assisted a class of 45 students on weekly lab exercises including Social Media Analytics and Natural Language Processing

• Researched various time series clustering methods and designed a framework using Time Series K-means to derive consistent archetypes from footfall datasets, obtaining remarkable results
• Performed spatial clustering to identify areas with similar footfall archetypes to capture key land uses contributing to footfall trends
• Trained a Name Entity Recognition (NER) model using spaCy to extract and geolocate spatial references from unstructured text
• Developed a Streamlit Application POC which integrates location NER model with OneMap location search API
• Presented projects to C-Suite and Professor Paul Cheung from National University of Singapore

• Devised 3 Tableau dashboards for a client project to generate business insights for better decision making
• Applied Machine Learning models in data analysis - Time Series Forecasting and Customer Segmentation Clustering

• Performed SafeEntry visualisations to generate insights into adoption and utilisation rates for essential and non-essential services
• Enhanced Sentiment Analysis Model on Supplier Reviews with Natural Language Processing techniques and Ensemble Learning to achieve better prediction accuracy
• Operationalised a Fraud Detection model to derive a risk scoring for each company using Machine Learning and Anomaly Detection methods - DBSCAN Clustering and Regression

• Conducted academic consultations to a class of 40 students on various data structures and algorithms to solve problems

• Devised a business proposal to expand SISB business model for pre-school education
• Conducted strategic analysis on pre-school education landscape using frameworks - Porter’s 5 Forces, SWOT, PESTEL