A web application which accesses 10+ AWS tools to provide users with recommendations on nearby places & spots based on their ratings and preferences. It also supports functions including text message notification, voice control, smart album, etc.

System Architecture

System User Interface

My Contribution

·System Design: Initialized the prototyping approach for system structure design and implementation process.

·Front-end: Drafted UI prototype using MockPlus. Implemented graphical interfaces using jQuery. Implemented Cognito for user authentication. Implemented data-flow related functions with APIGateway SDK.

·Back-end: Initialized APIGateway. Built Spark clusters on EC2 with EMR. Utilized various tools to implement functions, such as Lambda for server-less service, S3 for data & front-end storage, DynamoDB for database management, Lex for NLP, Rekognition for image classification, SNS & SQS for text message service and ElasticSearch for data searching. Accessed Google Map/Places API for location, image and social rating & reviews resources. Also implemented a VPC for application security.

·System Test: Completed unit tests & integration tests, participated in troubleshooting and system performance improvement.

Github Link

Youtube Link

A Big Data Analysis on Yelp Dataset Using Spark, NLTK, Machine Learning, Plotly, and High Performance Computing (2019).

My Contribution

·Configured HDFS and HPC cluster to store Yelp raw data, run scripts and maintain big data environment.

·Implemented and executed algorithms with PySpark to retrieve the distribution of bars in NA, regions with most reviews or active users, and locations of some “small but fine” restaurants.

·Retrieved the most popular keywords in public comments of certain types of restaurants using NLTK library.

·Implemented machine learning model with logistic regression using Scikit-Learn library to predict if a restaurant is likely to survive in the next 6 years. After optimization, the final prediction accuracy reached 90.6%.

·Created practical value by calculating the most critical negative and positive factors to a restaurant that had survived for 6 years.

Github Link

PDF Report