Image generation and transformation with GAN
Develop two generative adversarial networks with the ability to generate digits and transform pictures in one domain to another domain using conditional-GAN.
(Keras, GAN, AWS, EC2, MINIST, C-GAN, Computer Vision, Convolutional Neural Networks)
Hospital Return Rate Prediction
As a member of analytics team, I utilize patient
information to predict whether a patient will return to a hospital within 30 days of being discharged.
With return prediction information, hospital or emergency center could prepare in advance to meet
further room requirements and make recommendations to alleviate overcrowding.
(R, Healthcare Industry, Classification Prediction Modeling, Logistics Regression, Random Forest, Ensemble Methods)
Movie Industry Development Analysis
As a analytics team, we analyze the development information of movie industry in the past 100 years. Our dataset includes movie basic information and its relavant rotten tomatoes scores information. Our goal is to provide valuable insights for movie producer newbies on how movie industry changed and developed through the past 100 years and what factors might affect movie's quality, helping new movie producers to be more successful on the start of their career.
(Jupyter Notebook, Python, NumPy, Pandas, Matplotlib, Seaborn, Visualization)
Customer Relationship Database Management System
A database management system for HVAC Mechanics LLC. The system allows the company to efficiently manage the customer, supplier, service, transaction and requests information, improving the efficiency and enhance the management of HVAC.
(SQL, LucidChart, Microsoft sql server management studio, Database management, Relational Database)
Data Science for Good: City of Los Angeles
The content, tone, and format of job bulletins can influence the quality of the applicant pool. Overly-specific job requirements may discourage diversity. The Los Angeles Mayor’s Office wants to reimagine the city’s job bulletins by using text analysis to identify needed improvements.
The goal is to (1) identify language that can negatively bias the pool of applicants; (2) improve the diversity and quality of the applicant pool; and/or (3) make it easier to determine which promotions are available to employees in each job class.
(Jupyter Notebook, Text Analysis, Natural Language Processing, Regular Expression)
Stock Prediction in Supply Chain Industry
After analyzing the problem, I found the reason that delivery estimated accuracy is underperforming is because of the mislabeled items. In general, the buyable attribute of the ASINs are updated by in-stock managers manually, which is time-consuming and inefficient. With the development of big data, machine learning modelling technics are widely utilized in the IT industry to improve business performance and work efficiency.
(R, Binary Classification Prediction, Supply Chain Industry, Random Forest)