Raffael Yuliang Gao

高宇梁 · Data Scientist · Consumer Insight

Years Exp.

DS Projects

Publications

Python XGBoost NLP Keras Tableau SQL R Time-Series Forecasting Statistical Modelling Clustering

A resourceful data scientist with 4+ years of quantitative market research at Kantar — translating complex consumer data into actionable insights for global brands.

Unlock the Power of Data & Statistics

About

I come from a unique intersection of biology, behavioural science, business strategy, and data science. Over four years at Kantar Worldpanel in Shanghai, I led strategic roadmaps and deep-dive research for global FMCG stakeholders — including Shiseido, Estée Lauder, and Beiersdorf.

I pioneered the "Fusion" methodology, integrating purchase panel data with custom surveys — a diagnostic framework later adopted firm-wide. My work has been presented at industry conferences and published as whitepapers that shaped post-COVID beauty recovery strategies.

Now pursuing a Cambridge Data Science certificate, I bring the rare combination of domain expertise, analytical rigour, and storytelling ability that transforms raw data into decisions.

Bilingual

Fluent in English and Mandarin. Experienced across North American and Asian markets.

Insight-Driven

Domain expertise in FMCG, beauty, and consumer behaviour. Data-to-decision translation.

Fast Learner

From biology to business to data science — I adapt quickly and deliver under pressure.

Skills

Languages & Frameworks

Python R SQL Scikit-learn Keras TensorFlow Pandas NumPy NLTK PySpark Matplotlib Seaborn Git

Machine Learning

XGBoost Random Forest Neural Networks CNN LSTM NLP Topic Modelling Time-Series Forecasting K-Means Clustering Hierarchical Clustering PCA t-SNE Anomaly Detection Feature Engineering Model Evaluation

Visualization & BI

Tableau Power BI Matplotlib Seaborn Plotly Excel PowerPoint SPSS Dashboard Design Data Storytelling

Domain Expertise

Consumer Segmentation Purchase Panel Data FMCG / Beauty Survey Design Perfomance Diagnostics Price Elasticity Brand Health Tracking Whitepaper Authorship Conference Speaking Stakeholder Consulting Bilingual EN / 中文

Projects

Multimodal Time-Series Forecasting

Neural network architectures combining CNN, LSTM, and dense layers for multivariate time-series forecasting on complex panel datasets.

LSTM SARIMA CNN Hybrid-Modelling Deep Learning Keras

View on GitHub

Unsupervised Learning: Consumer Clustering

K-Means and hierarchical clustering on purchase behaviour data to identify distinct consumer segments — driving targeted marketing strategies for FMCG brands.

K-Means PCA t-SNE Segmentation

View on GitHub

Anomaly Detection: Maritime Predictive Maintenance

Isolation Forest and PCA-based anomaly detection on maritime sensor data to flag equipment failures before they happen — reducing unplanned downtime.

Isolation Forest PCA Anomaly Detection Predictive Maintenance

View on GitHub

Anomaly Detection: Maritime Predictive Maintenance

Isolation Forest and PCA-based anomaly detection on maritime sensor data to flag equipment failures before they happen — reducing unplanned downtime.

Isolation Forest PCA Anomaly Detection Predictive Maintenance

View on GitHub

Applied NLP: Topic Modelling & Sentiment Analysis

Extracted customer sentiment themes from unstructured review data using BERTopic, LDA, and large language models. Delivered actionable product insights.

BERTopic LDA BERT Falcon-7b NLP

View on GitHub

Anomaly Detection: Maritime Predictive Maintenance

Isolation Forest and PCA-based anomaly detection on maritime sensor data to flag equipment failures before they happen — reducing unplanned downtime.

Isolation Forest PCA Anomaly Detection Predictive Maintenance

View on GitHub

Supervised Learning: Student Dropout Prediction

Multi-stage classification pipeline predicting student dropout risk using XGBoost and neural networks — achieving 0.89 AUC for early intervention timing.

XGBoost Decision Trees Neural Nets Classification

View on GitHub

MORE PROJECTS ·

MARKET RESEARCH ·

MORE PROJECTS · MARKET RESEARCH ·

Anomaly Detection: Maritime Predictive Maintenance

Isolation Forest and PCA-based anomaly detection on maritime sensor data to flag equipment failures before they happen — reducing unplanned downtime.

Isolation Forest PCA Anomaly Detection Predictive Maintenance

View on GitHub

Anomaly Detection: Maritime Predictive Maintenance

Isolation Forest and PCA-based anomaly detection on maritime sensor data to flag equipment failures before they happen — reducing unplanned downtime.

Isolation Forest PCA Anomaly Detection Predictive Maintenance

View on GitHub

Anomaly Detection: Maritime Predictive Maintenance

Isolation Forest and PCA-based anomaly detection on maritime sensor data to flag equipment failures before they happen — reducing unplanned downtime.

Isolation Forest PCA Anomaly Detection Predictive Maintenance

View on GitHub

Anomaly Detection: Maritime Predictive Maintenance

Isolation Forest and PCA-based anomaly detection on maritime sensor data to flag equipment failures before they happen — reducing unplanned downtime.

Isolation Forest PCA Anomaly Detection Predictive Maintenance

View on GitHub

Publications

Seminar Speaker

DEC 2024

Finding New Growth Area Under A New Era

China Cosmetic Annual Conference · Guangzhou, China

Seminar Speaker

AUG 2024

Building a more Meaingful 360 profile of Consumers

Kantar Worldpanel Annual Client Day · Shanghai, China

Seminar Speaker

AUG 2023

GUO-CHAO Evolution: The Past, Present, and Future

Kantar Worldpanel Annual Client Day · Shanghai, China

Whitepaper Author

MAR 2023

China Beauty Market Post-COVID Recovery Whitepaper

Published by KANTAR Worldpanel · Shanghai, China

Experience

Sep 2025 — May 2026

Certificate. Data Science with Machine Learning

University of Cambridge (FourthRev) | Online

Supervised & unsupervised learning, neural networks, NLP, time-series forecasting, feature engineering, and model evaluation.

Python Keras Scikit-learn NLP

May 2021 — Mar 2025

Associate Market Research Manager

Kantar Worldpanel · Shanghai, China

Led strategic roadmaps for global beauty clients. Pioneered "Fusion" methodology integrating purchase panel data with custom surveys — adopted as a firm-wide diagnostic standard. Authored the Post-COVID Beauty Recovery Whitepaper and presented at industry conferences in 2023–2024.

Consumer Insights Segmentation Panel Data R

Sep 2019 — Aug 2020

M.S. International Business

Hult International Business School · London, United Kingdom

Business strategy, marketing analytics, and cross-cultural management. Honed the strategic framing that bridges quantitative analysis with executive storytelling.

Sep 2017 — Oct 2018

PG.Dip. Human Resources Management

Western University · Ontario, Canada

Organisational behaviour and strategic human resources planning. Including a four month practiccum as human resources assistant at TSC Stores L.P. located in London Ontario Canada.

Sep 2013 — Apr 2017

B.Sc. Honours Biology

Western University · Ontario, Canada

Foundational training in scientific method, statistics, and quantitative reasoning — the analytical backbone as well as critical thinking capabilities. President, Calligraphers of Western, 2016. VP Promotion, The Acapella Project, 2017.