Hands-on with Business Analytics
55 students in 14 teams
Business Analytics is the growing, inter-disciplinary field of bringing data to build business insights and support decisions. The goal of the course is to bridge the divide between technical skills and business know-how. Students self-propose projects across a variety of industries and interests. Projects are focused on providing prescriptive or causal analysis, in addition to predictive and descriptive analysis.
The project will explore the relationship between disasters worldwide and its impact to the tourism industry within region and outside the region. Through passenger inflow data and disaster data, we will test the existing assumptions and belief that countries with more disasters would naturally be riskier and less attractive to tourists. Disaster data will include natural and man-made disasters which has potential transformative impact and also strive to find correlations between certain tourism amongst neighbouring countries or continents.
Kickstarter is a platform which raises fund and help bring creative projects to life. So far, 103 K projects have been successfully funded with a total funding of 2 billion USD. We plan to study what is the driving force behind the project to be successfully funded. We are doing a deep dive analysis to find out the driving factors and are incorporating Google trends to study the same. Google trends will act as a proxy from popularity, sentiments & the public opinion/ interest about the projects. They will supplement the already existing information about content, location, categories etc. Any information backed by data will help the artists to design, position and implement the projects so that they have higher chances to succeed in future. For the website, they may improve their services by adopting a more consultative approach and helping to onboard the new entrants based on the past data study.
Gun control is one of the most contentious issues in United States. With every incident of mass shooting, the debate over gun control reignites. On one side, it is argued that increase in gun ownership will cause more disputes to result in a homicide. On the other hand, if potential victims carry guns, it can act as a deterrent to violent crimes. The purpose of this study is to enable a data driven analysis of this issue rather than being driven by political ideologies or emotions. Though the positive correlation between guns and crime is widely known, we aim to study whether there is any causal impact of guns on crime. We address the challenges encountered in a causal study such as reverse causality , omitted variable bias , mis-measurement of independent variable (proxy) by using Instrument Variables and Regression Discontinuity approach.
This project aims to help the lenders on Kiva to have a better assessment of the risk involved in each loan application, so to understand why some loans have defaulted and also to identify the root causes and causal relationship with external factors such as the economy, weather, etc.
Bike-share programs in cities have enjoyed growing popularity in recent years. From just 13 programs in 2004 it has grown to 855 in 2015. However, doubts remain on their effectiveness on reducing traffic congestion. According to INRIX, New York City ranks #5 in the Top 10 worst traffic cities in America. Using transactional data on taxi-ridership and Citi Bike data, we hope to understand if the introduction of Citi Bikes (in 2013) reduced congestion in New York City and its impact on taxi ridership. The insights can inform City Planners on decisions relating to expanding the bike-share program further, as well as help the program operator identify untapped customer segments that travel on popular bike routes via taxis.
Associating light with safety has been universal. In most people's minds, there is a simple and direct relationship between lighting and crime i.e. better lighting will reduce crime. On the other hand it can be argued that street lights might enable criminals as much as they do their potential victims. We aim to study City of Chicago crime and street lights data and analyze if there is any causal impact of street and alley lights on crime in Chicago.
Tobacco use is the leading preventable cause of death and disease in Canada. With smokers starting at a very young age, what causes people to start smoking? With data science, we try to find the leading causes of people getting addicted to tobacco use. this will allow the Government to formulate the right policies for the right demographics of smokers. From an array of data sources, we try to cover all the reasons leading to smoking habits so that there can be better and effective measures to combat the issue.
When disaster strikes close to home, the tragedy is experienced first-hand by the locals. Our project investigates if such experience in close proximity (i.e. home country) will shake the confidence of the affected individuals to travel destinations that are unaffected but potentially susceptible to similar disasters. For hotel providers where 1 or 2 nationalities are main revenue drivers, a good knowledge of how specific disasters in their core customers’ home country may impact their visitorship allows them to take early actions to counteract changes in demand. Using the Japan 2011 tsunami as the focal point, we investigate if the tsunami has any impact on the travel patterns of Japanese tourists to unaffected beach resorts in the period after.
Just stand at a street corner and hold your arm straight out. It’s pretty much all you need to do to hail a taxi in New York City. Tipping the cab driver is customary in New York, but not mandatory. However, it is considered to be civil to give the driver at least a 10%, 15% or 20% tip if he got you to your destination safe & sound, or perhaps helped you load your baggage or showed you the best steak place in the Big Apple. In this study, we analyze the effect of uncommon disruptions(like large-scale gas explosions, subway breakdowns) which shook New York City and its impact over taxi tipping behavior. Do cabbies take undue advantage of the situation and charge riders more than usual? Or, are riders thankful enough to tip the cabbie a little extra? The insights from our analysis would help not only the Taxi & Limousine Corporation in NYC to better manage the demand for cabs but would also suggest appropriate tips to be given.
During this project, we try to perform analytics on the significance of impacts that various types of weather have on the business check-in of various types of physical stores located at New York City. We have acquired datasets of business checkin as well as daily weather from Foursquare, Yelp and NOAA. By analyzing the data, we try to differentiate the significance level of different types of weather (Sunny, Drizzle, Heavy rain, Snow, Fog etc.) impacts on various business types such as Bars, Coffee shop, American Restaurants, Gyms and so on. Furthermore, we also perform a causal analysis on the case of Hurricane Sandy, which unexpectedly occurred on Oct 28-29, 2012 and had a major impact on the region of New York City. We try to apply Difference-in-difference model to analyze the causal relationship between the storm and the business visits of physical stores in New York City.
Not sure how to select a car that best meets your needs? Applying the Analytic Hierarchy Process (AHP), the team aims to recommend prospective car buyers the most suitable car model and drive type based on their considerations in various aspects such as cost, safety, environmental friendliness, speed, capacity etc, from our database of more than 3,800 drive types across 11 brands and (approximately) 80 models.
Our project is inspired by a famous American photographer Bill Cunningham, whose daily job is to shoot the styles on the street or on big events and then distill the latest trends from the runways of Paris to the colorful streets of New York. Discovering fashion trends is considered especially crucial to some fast fashion brands like H&M, Forever 21, Zara etc. These brands always need inspiration from all kinds of resources to fulfill the ever-changing demands. Our project aims at automating Bill Cunningham's work by doing clustering on a huge dataset of fashion photos, summarizing the fashion trends, and constructing a web dashboard to constantly update the trends.
The total number of restaurants in the USA is now over 616 thousands and keeps an increase of 7 percent per year. Restaurants, as a typical and important business tightly connected with people’s daily life, are often treated as an indicator of local economic scale and resident income, reflecting people’s quality of life in that region. However, there are no previous studies unveiling the real factors that influence the number of restaurants in a region during the last few decades when the total number of restaurants in the USA ever experienced the greatest growth. In this project, we collect data from the largest restaurant review website in the USA named “Yelp” and a demographic information website "City-Data", make use of statistical models and thereby summarize the result obtained from statistical analysis tools. Eight factors have been proved to be the significant variables that have a linear relationship with the number of restaurants in an area defined by ZIP code. We interpret these eight factors under the social circumstance of the USA and analyse the reason behind. Last but not least, some suggestions have been provided to those people who plan to open a restaurant in USA to help them make a decision on location choosing.
Is it possible to accurately forecast attendance at a sporting event? The Melbourne Cricket Ground (MCG) is a sporting ground in Australia, with capacity of 100,000 people that primarily hosts Cricket in summer and Australian Rules Football (AFL) during winter. The MCG in partnership with students from the MSBA & MOC programmes is working on a project to more accurately forecast attendance at their events. The consequences of an inaccurate forecast are significant. When underestimated, it leads to a shortage of staff to provide services such as security, food & beverage. When overestimated, it leads to an unnecessary amount of resourcing being provided to attend to guests. Focussing on the AFL season, the team is building a regression model based on data from 2007 to 2014 and testing this model against attendance from the 2015 season. At STePS, we will present our model and results.