(WIP) KPI & AB test analysis dashboard

A user engagement dashboard with KPI, data analysis w/ slicing, and AB test analysis capability by implementing E2E from synthetic data creation to visualization.

Mental health risk classifier using LLM

Based on user text, detect if a user has particular mental health risk by different set of classification tasks.

How are you, America?

Uncovering mental health states and trends using data visualization and story telling.

Large scale flight delay prediction

Predicting if a departing flight will be delayed or not using 30M+ rows of historical flight data using Databricks and pyspark for large-scale processing.

Dashboarding in 6 different tools

Comparing 6 dashboarding library/tools and see how they compare, from Excel to Tableau to Plotly

Classifying text with only POS data using LLM

Project to classify text written by neurodivergent people by only using Part-of-Speech (POS) data extracted from text in order to respect privacy.

Quantitative analysis to words frequently used by ChatGPT-as-an-editor

Prompted ChatGPT to edit data science articles on Medium and analyzed the differences in word frequency before and after the ChatGPT edits. Additionally analyzed the word frequency change between 2022 (pre-chatGPT) and 2023 (post-chatGPT) to find similar trend in word frequency change.

Correlation analysis of world happiness report data with other open data

Deeper dive of the World Happiness Report (https://worldhappiness.report/) data with other open data from the World Bank (e.g. income, health expenditure) to see what KPIs/signals correlates to people's happiness.

Causal impact analysis of marriage to mental health

Analysis of how marriage effects the chances of having depression based on NLS (National Longitudinal Survey) data. Analysis was done using traditional methods (simple logistic regression, PSM, IPW) and newer methods (t-learner, double machine learning) w/ DoWhy & EconML. Results stored in github.