Data Science Machine Learning Statistical Modelling

Yihang Feng

MSc Data Science student at LSE; BEcon in Economic Statistics from HUST.

I build forecasting and causal-inference models, and the software tools around them.

Methods

Forecasting
Random Forest, LSTM, GRU, TCN
Causal inference
DID, PSM, IV-2SLS, spatial panels
Software
Python, Stata, Swift, JavaScript

Research

Machine Learning Models for Gold Price Prediction: A Comparative Study

Compared Random Forest, LSTM, GRU and TCN models for forecasting gold prices from historical financial data and global market indicators.

TCN: best accuracy.
GRU: best efficiency.

  • Time series
  • Deep learning
  • Finance

Impact of the Carbon Emission Reduction Facility on Clean Energy Development

Led an empirical study of how China’s Carbon Emission Reduction Facility affects clean-energy development, financing barriers, innovation and sustainability outcomes, using Python and Stata workflows that address missing data, selection bias, endogeneity and robustness.

  • DID
  • PSM
  • IV-2SLS
  • MICE
  • Spatial panel

Comparative Analysis for Carbon Monoxide Prediction in Air Quality Forecasting

A carbon-monoxide forecasting project for air-quality prediction, delivered as a research paper and a final presentation.

  • Forecasting
  • Air quality

Local Healthcare Systems in Shaanxi

Led a ten-student team from economics, public health and computer science in Weinan, coordinating interviews and site visits with government and medical institutions, and collecting and validating more than 1,200 resident questionnaires.

  • Survey data
  • Interviews

Projects

Slipstream

A reading aid for Chinese-speaking readers of English mathematics papers and textbooks. Select a word, passage, formula or small figure region to see Chinese beside the source, ask for a contextual explanation, and save useful concepts as local Markdown cards.

Signed and notarized
for macOS.

  • Electron
  • React
  • KaTeX
  • Swift

Chronicle

A fully offline, automatic work log for macOS. It turns foreground-app activity into reviewable work blocks, so a day or week can be reconstructed without relying on memory, and exports to Markdown and CSV.

Encrypted on-device.
Nothing is uploaded.

  • Swift
  • SwiftUI
  • AppKit
  • SQLCipher

Notes on Economic Statistics

An open knowledge base of mathematics, economics and computer-science course notes, organised as linked concept notes with their sources and meant to be read in Obsidian.

  • Markdown
  • Obsidian

Education

  1. Expected 2027

    MSc Data Science

    London School of Economics and Political Science London

  2. 2022 – 2026

    BEcon, Economic Statistics

    Huazhong University of Science and Technology Wuhan

    GPA 4.14 / 5.0 (average 86.5 / 100). Coursework includes Multivariate Statistical Analysis, Econometrics, Stochastic Processes, Game Theory and Advanced Programming Language.

  3. 2025

    Summer School in Artificial Intelligence and Machine Learning

    St Antony’s College, University of Oxford Grade A−

Honours
Honorable Mention, Mathematical Contest in Modeling (MCM), 2024
Languages
Mandarin (native); English (IELTS Academic 7.0)
Music
Diploma-level piano performance, China Conservatory of Music grading examination; solo concert, 2023

Contact

For research, data-science roles or collaboration, email is the best way to reach me.