Duong's portfolio

Data Science, Machine Learning & AI

Demand Prediction Model with SHAP Interpretation

The objectives of the analysis are: Understand demand across different price ranges, brands, shops, and seasons (1), Build a predictive model to forecast demand (2), and Explain observations and attributed variables importance in the model (3).

Tools used: Python, RandomForestRegressor, TreeExplainer, SHAP

Text Classification Model with BERT and SHAP

The objectives of the analysis are: Understand demand across different price ranges, brands, shops, and seasons (1), Build a predictive model to forecast demand (2), and Explain observations and attributed variables importance in the model (3).

Tools used: Python, BERT, transformer tokenizer, SHAP

Business Intelligence & Report

Each data collected by business can make a significant contribution to its value; however, this is not universally utilized. Business intelligence (BI) is a vital tool for organizations seeking to transform raw data into actionable insights. Depending on the business needs, the insights could be related to sales, financials, operations, safety measures, and so on.

This example report highlights key performance indicators (KPIs) and explores future sales prospects of Wish platform in August 2020. These visuals, created in PowerBI, enable efficient monitoring of trends and support data-driven decision-making.

Tools used: Power BI

Focus area: KPI tracking, sales report, data storytelling

Projects

Financial analysis - ESG and CFP

This project is part of my analysis in my Bachelor's thesis, which consisted of literature and different hypotheses regression analysis of ESG and different firm characteristics on stock prices as an indicator of financial performance. The whole process includes extensive data collection, cleansing, dimensionality reduction, and regression on Excel.

Read the full version from the Aalto University repository

Hotel room price prediction - Bayesian approach

In the Bayesian Data Analysis course, we, Duong and Vee, applied the knowledge acquired to build predictive models for hotel prices using R. The dataset was extracted from R with information on historical prices and booking details. By approaching with both non-hierarchical and hierarchical Bayesian models, we help optimize price predictions for managerial implications and for travelers to understand the price drivers.

Read the brief presentation embedded below: