-
Xiamiu - Online Music Database
- A web application to help music fans manage their favorite music and explore their musical interests.
- Visit
-
Awesome Blog
- An online blog for users to post and share their articles about data structures and algorithms.
- Visit
-
Hunt the Wumpus
- Hunt the Wumpus is an adventure game developed by Gregory Yob in 1973. In the game, the player moves through a series of connected caves, arranged as the vertices of a dodecahedron, as they hunt a monster named the Wumpus. I design a Hunt the Wumpus game using JAVA with mvc design pattern. It could be played in both text-based mode and GUI based mode(graphical user interface is built using Java Swing).
- View
-
Google Analytics Customer Revenue Prediction
- This project aims to predict the revenue that consumers buy after they entered into the google merchandise website.
- View
-
E-commerce Recommender System
-
• Built main database with MongoDB and built cache database with Redis for low-latency queries in real-time
• Implemented the offline recommender service by using ALS algorithm in Spark MLlib to train the latent factor model and calculating the recommendation for user based on user ratings data in MongoDB
• Implemented the real-time recommender service with Flume(collecting log data), Kafka(buffering unprocessed data) and Spark Streaming(stream processing), and then updated the result to MongoDB - View
-
• Built main database with MongoDB and built cache database with Redis for low-latency queries in real-time
-
World Trade Web Visualization
- This is one synthetic RShiny application which shows the trade relationship between countries over the world and the trade trend of each country. To reach this goal, firstly I design the structure of this application. It mainly includes four parts: global.R, server.R, ui.R and helpers.R.
- View
-
Movie Recommender System
- Predicted ratings for movies with Tensorflow Content-based filtering and Spark Collaborative filtering, then recommended movies to users
- Try it
-
GitHub Commits Sentiment Analysis
-
• Measured emotions expressed in different projects’ commit comments by using VADER, one lexical sentiment analysis method, wrote a module by using sqlite3 to connect to SQLite and run query in Python
• Built one analyzer using numpy, pandas, texttable and so on in Python to study the relationship between the expressed emotions and different factors. - View
-
• Measured emotions expressed in different projects’ commit comments by using VADER, one lexical sentiment analysis method, wrote a module by using sqlite3 to connect to SQLite and run query in Python
-
Adult Income Prediction
-
Implemented 3 versions of tree-based models and neural network to achieve our goal in classifying whether an individual’s annual salary is more than $50,000.
Estimated imbalanced data with SMOTE technique and improved accuracy of all models to 86% and higher - View
-
Implemented 3 versions of tree-based models and neural network to achieve our goal in classifying whether an individual’s annual salary is more than $50,000.
-
User Behavior Analytics
-
• Created Python based web crawler engine and collected the music libraries and user information of samples of more than three million users
• Designed MySQL database to manage 2GB data, generated six relational tables under 3NF
• Provided Ximai user profiles with fifteen features such as socioeconomics and demographics
• Evaluated Influence of users by using H-index method - JS View
-
• Created Python based web crawler engine and collected the music libraries and user information of samples of more than three million users
-
Diversity of Music Tastes
- The diversity of people's musical tastes is one of the significant parts which helps people to better understand the behavior trends and cultural preferences of people. In this paper, based on Hill-type true diversity, we propose an improved diversity metric that fairly captures the diversity of musical tastes.
- View Paper
-
World Trade Web
- The backbone of world economy system is xonstructed by the World Trade Web. Here we identified influential countries in this network through analyzing 20 year's global export and import trade data, employing a weighted DHC theorem on it. The results suggested our method to be a good indicator when qualifying node influence in directed and weighted networks.
- JS View Paper
