Dasen Ye
Data Science and Business Analytics
Cohort 2024
HEC Montréal
Data Science and Business Analytics
Cohort 2024
HEC Montréal
Data Science and Business Analytics
HEC Montréal
Master's
I am currently enrolled in the MSc. Data Science and Business Analytics program at HEC Montreal with a GPA of 4.0/4.3. The relevant courses that I took/am taking include Machine Learning, Statistical Modeling, Statistical Learning, Advanced Statistical Learning, and Forecasting Methods. Previously, I obtained my Bachelor's degree in Mathematics and Physics and a Master's degree in Physics (with the research area being theoretical high-energy nuclear physics) from McGill University.
Greetings! My name is Dasen, currently pursuing my Master's degree in Data Science and Business Analytics at HEC Montreal, alongside participating in the Fin-ML CREATE Program. As a Master’s student in data science, I am a good problem-solver with solid theoretical and practical skills. Proficient in Python and R, as well as being familiar with advanced mathematical and statistical concepts, I possess the ability to tackle complex challenges and provide innovative solutions and insights.
Change-Point Prediction of Financial Variable with Textual Sentiment and Macroeconomic Data
We examine textual market sentiment data and traditional macroeconomic data's ability to predict change points in several financial variables. Our results show that textual sentiment data better predict change points for topics related to the national GDP, job market, real estate, and retail than the macroeconomic data while having significantly worse performance on price & interest rate and surveys data than the latter.
Enhancing Industrial Production Forecasting with Machine Learning and Sentiment Analysis
Accurate forecasting of industrial production is critical for economic policy, investment decision-making, and risk management. Traditional econometric models often struggle to capture economic data's complex, non-linear relationships. This study explores the potential of machine learning techniques, particularly elastic net, gradient boosting, random forests, multilayer perceptrons, and recurrent neural networks (RNNs), to improve industrial production forecasting by integrating macroeconomic, financial, and sentiment-based data. Using a dataset covering U.S. industrial production from 1994 to 2017, we examine the predictive performance of these models, both with and without sentiment data. Our findings indicate that traditional models, including elastic net and random forest, provide strong baseline predictions, while deep learning models, particularly RNNs, benefit significantly from including sentiment information. These results underscore the value of incorporating sentiment analysis into economic forecasting and highlight the strengths and limitations of different machine-learning approaches in predicting industrial production.
