Business Data Insights in the NYC Coworking Industry
The skills the author demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
For our final capstone project, we utilized R & Python to develop key metrics & data analytics tools for a start-up in the co-working industry. They partner with restaurants & hotels in the NYC area to provide an alternative working & meeting environment for freelancers, businesses, and telecommuters.
Note: proprietary details will not be shared in this blog post to protect the company's information.
EXPLORING THE DATA
The core data for the company lies within their database, as well as supplemental spreadsheets & data collected by 3rd party platforms. The information was joined & categorized into Users, Locations, Payments, Activity & Insights. The datasets ranged from 20 to 200K observations, with anywhere from 2 to 20 features.
The objective is to understand what the typical customer looks like. There are several ways to divide the users into groups; differences between B2B vs. B2C, Paying vs Trialing, and Active vs. Inactive were investigated. Additional information about the user was collected through surveys & self-filled bios, and other details such as gender & USA state were supposed from extracting first names & phone number area codes.
The objective is threefold; to examine how many users frequent which location, to examine how long each session is for each location, and to examine how many weeks the average user checked-in per location. This allows for easy comparison in terms of popularity and value-to-customer each location brings.
The objective is to visualize how many new users were added compared to how many users were ending their subscription. This was displayed with time series graphs on both a weekly & monthly basis. Certain trends could be seen correlating with seasonal changes or effective advertising campaigns.
The objective is to review textual feedback from complaints, cancellation reasons & surveys and gain some insights on new features or fixes the company could employ to improve customer satisfaction. This portion was completed using Python NLP libraries to get word counts of 1-, 2- & 3- n-grams as well as bar graphs displaying the most common feedback per location.
The objective is to analyze financial ROI by comparing cost spending vs. subscription revenue, drilled down into yearly segments. It is also to look into other factors such as failed payments by users, spending categories & discount coupons affecting the net gain. Since we did not have access to some of the spending information, the framework was created with pseudo-values.
A huge thank you to the company for letting us work with their live data, challenging us to use our creativity to focus on real KPIs that affect their business. As the company rapidly changes (and grows), the importance of data collection & analysis will be even more critical to make future decisions.