Skip to content

Jasmine-Qu/Course_Project_Collection

Repository files navigation

Project_Collection

Project1-Causal-Inference

Topic: Estimation of Demand Function for Train Travel

Project group member: Chen Yiqiu, Chen Keyi, Guo Lu, Mi Jiale, Qu Mingyu

In this project, we estimated the demand function of train travel by using data on train ticket sales at a particular train station. Treating the demand and supply equations of train travel as a system of two simultaneous equations, we adopted Two-Stage least squares (2SLS) regression as the primary analytical technique in modeling the demand function. A selfengineered instrumental variable, num_days_ahead, is particularly chosen to solve the endogeneity issue in demand analysis.

Project2-Policy-Impact-Analysis

Topic: Estimating the Announcement Effect of the Volcker Rule

Project group member: Chen Yiqiu, Chen Keyi, Guo Lu, Mi Jiale, Qu Mingyu

In this project we estimated the announcement effect of Volcker Rule (new banking regulation in US) on US banks. More specifically, we analyzed: (i) Did the banks decrease their trading assets after the announcement of the new regulation? (ii) If they responded to the regulation, which banks responded most and which banks least? Why? (iii) Remember robustness, and how should banks or regulators use these results?

Project3-Muti-armed-Bandit

Topic: Determining the best combination

Project group member: Chen Yiqiu, Chen Keyi, Guo Lu, Mi Jiale, Qu Mingyu

This project is based on one simple experiment designed to encourage visitor to Obama’s selection website to sign up and donate, which was conducted by Obama’s fundraising campaign. Though this experiment, Obama’s team selected the bestperforming combination of button and media and increased click-through rate (CTR) as well as donations. Our project is a simple duplicate of this experiment with 24 different arms. Our task is to maximize the number of visitors who sign up to the website. It is essentially a multi-armed bandit model (MAB) with bandit feedback, and we try to identify the best arm by algorithms to get the highest total reward.

Project4-House-Price-Forecast

Topic: Forecast house price in Singapore

Project group member: Kwok Shi Ann Sheranne, Tan Lipin, Zhang Yixuan, Qu Mingyu

Daniel is a young professional, who is currently considering purchasing either a HDB unit or private property (e.g., apartment/condo) for a 3-year investment. In this project, we aim to develop a predictive model for candidate site selection to help Daniel understand SG property market and maximize his return on investment, i.e., % gain in 3 years.

Project5-Vehicle-Demand-Forecast

Topic: Integrated Vehicle Pre-allocation Model

Project group member: Kan Yan Zhe, Li Xinlin, Liu Xinyu, Qu Mingyu

One pressing issue faced by taxi operators is the over-supply in downtown area and the insufficient demand within its surrounding areas during peak hours. Our report aimed to help the operators proactively allocate their idle vehicles from downtown to the demanding areas. We selected both the linear ARIMA hybrid model and regression tree as the baseline to predict region demands using weather information, and to optimize the dispatch strategy under the network flow problem structure by utilizing the predicted demands.

Project6-Football-Analytics

Topic: Football Analytics-Build the Dream Squad

Project group member: Hao Ruoxin, Nicole Lee Zhi Ying, Liu Xinyu, Zhang Yuxuan, Qu Mingyu

This project is in the shoes of football club managers and aims to use both on-field and off-field statistics to build a dream squad consisting of the most competent players to date. The business task can be viewed as a candidate-position assignment problem in the field of optimization research. We solved the optimization problem under two different scenarios – with an unlimited budget and with a budget constraint respectively. To give an accurate estimation of key parameters in the optimization formula such as player market value and the cost of position transition, we will first predict the most suitable position for a certain player using classification and then forecast his expected wage using regression. Finally, with the predicted values, we will optimise the objective function with corresponding constraints to find our dream squad.

Project7-Predict-Online-News-Popularity

Topic: Predict Online News Popularity

Project group member: Chen Yiqiu, Hao Ruoxin, Lin Yijiang, Mi Jiale, Qu Mingyu

This project is about prediction popularity of an online news, i.e., to predict the popularity based on the number of shares in social networks by using machine learning models. We define it as a multi-class classification problem.

Project8-Bank-Marketing

Topic: Bank Marketing for Term Deposit Subscriptin

Project group member: Chen Yiqiu, Hu Xinjie, Lee Hui Gek Judy, Wang Huijuan, Qu Mingyu

In this project, we aim to develop a classification model for a Portuguese bank on which customers contacted by the bank’s telemarketers would subscribe to the bank’s term deposit products.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published