Yannan Su
Data Enthusiast, Computational Neuroscientist
More About Me: LinkedIn|GoogleScholar|Digital Garden
In my spare time, I take photos, read books , and write blog in Chinese.
Portfolio
Selected Projects
Research Projects - Visual Information Processing
Psychometrics | Bayesian Models | Mathematical Approaches
To gain insights how visual perception works as an “unconscious inference” and how our brain uses computational representations to process visual information, I measure visual percepts of humans and reveal the underlying models.
Learn more about my research projects
Daily progress in my PhD journey
Data Visualization | Interactive Dashboard
Approaching the end of my PhD, I set up a daily standup meeting and recorded my daily progress in a spreadsheet. This app is to visualize the data and help me to track my progress.
View dashboard app or View code on Github
BCG Virtual Experience Program - Customer Churn Prediction
Business Understanding and Strategies | Classification
In this open-access BCG Virtual Experience Program with Forage, I built a predictive model that can identify customers at high risk of churn for a utility company.
Diabetes health indicator
Classification | Model Tuning
Early diagnosis of diabetes is important to prevent the onset of complications. From survey data, I found health indicators that are most associated with diabetes.
Music effects on mental health
Exploratory Data Analysis | Correlation
How music affects our mental health? In this notebook, I used survey data to explore participants’ music listening habits and mental health status.
All Projects
Symptoms to diagnosis using NLP
NLP | Word Cloud | Neural Network | TensorFlow
NLP seems becoming powerful in the field of healthcare. In this project, I explore how to infer disease from symptoms using language processing and neural networks.
Spotify music recommender
Recommender System | Dimensionality Reduction | Streamlit App
I used Spotify’s audio features to build a music recommender system. I further developed a web app in which the recommender system can be customized by selecting the features.
View code on Kaggle or View App
Rental offers in Munich
Geodata | Regression
Hunting for a new apartment in Munich? I analyzed the data from Immobilienscout24 to find what factors affect the price of a rental offer in Munich.
A Starter for Forecasting Time Series
Time Series | Forecasting
Which methods are suitable for forecasting time series? I made a starter notebook to explore different models and find the best method to predict sales data.
Introduction to Probabilistic Programming with NumPyro
Bayesian | NumPyro
Probabilistic programming is a powerful approach to statistical modeling. In this tutorial, I introduced the basics of probabilistic programming with NumPyro.
Forest Cover Type - Compare Classification Algorithms
Classification | Model Tuning
In this project, I predicted the forest cover type (the predominant kind of tree cover) from strictly cartographic variables (as opposed to remotely sensed data). I compared different classification models and applied hyperparameter tuning to the best model.
Catch the CATs - Explore Feature Engineering
Exploratory Data Analysis | Feature Engineering
In this project, I performed feature engineering on various features, including binary features, low- and high-cardinality nominal features, low- and high-cardinality ordinal features, and (potentially) cyclical features.