DAA 7030 Module 3 Midterm Paper: Collecting, Mining and Presenting Public Data on Where to Open the Next Studio Example

From Two Public Data Sets to a Map of Opportunity: Data Mining and Visualization for a Studio Expansion

Student Name

MBA in Data Analytics, Thomas Edison State University

DAA-7030: Data Analytics and Visualization with Capstone Project

Mentor Name

Month Day, Year

From Two Public Data Sets to a Map of Opportunity: Data Mining and Visualization for a Studio Expansion

Purpose

The owners want to open a seventh studio within two years. Their instinct is to open in the next town over. This paper uses public data to ask a broader question: which ZIP codes in Bergen and Passaic counties combine the customers our studios attract with few existing competitors? Wedel and Kannan (2016) argue that marketing analytics now draws on rich data from many sources to support decisions across the marketing mix, including where to locate and whom to target, and that its value lies in turning data into decisions.

Collecting the Data

Two public data sets were collected for 70 ZIP Code Tabulation Areas.

Both were downloaded as tables, cleaned and joined by ZIP code into one csv file of 70 rows. Cleaning removed two ZIP codes with fewer than 1,000 residents and one post office box ZIP without population. Two derived measures were added: the share of residents aged 25 to 44, our core member age, and fitness establishments per 10,000 residents, a measure of competition.

Mining the Data

Loshin (2012) describes data mining as using algorithms to find patterns not obvious from reports. To group ZIP codes with similar profiles, I used k-means clustering on four standardized variables: median income, share aged 25 to 44, population and fitness establishments per 10,000 residents. Jain (2010) reviews fifty years of clustering research and notes that k-means remains widely used because it is simple and efficient, but that it requires choosing the number of clusters and assumes roughly spherical groups of similar size. I tested three to six clusters and chose four, the point after which additional clusters added little explained variation and the groups still made business sense.

Results

Our six existing studios all sit in Cluster 1. Cluster 2, eleven ZIP codes with high incomes, a larger share of residents in our core age group and few fitness centers, is the most promising. Three of its ZIP codes lie within twenty minutes of our main studio, which would allow shared instructors.

Presenting the Results

The owners will see two visuals. A choropleth map of the 70 ZIP codes, colored by cluster, shows at a glance that Cluster 2 forms a band along the Route 17 and Route 4 corridors. A scatterplot places each ZIP code by median income on one axis and fitness centers per 10,000 residents on the other, with point size showing the number of residents aged 25 to 44; the three recommended ZIP codes are labeled. Both visuals use position and color on common scales, which viewers read more accurately than pie charts or three-dimensional effects.

How the Process Generalizes

The same three-stage process, collect, mine and present, can be reused for other decisions. Adding the chain's own booking data to the public data, for example, would let us cluster neighborhoods by how many of their residents already take our classes, a measure of brand reach. Repeating the analysis each year as new Census estimates are released would show whether underserved areas are filling with competitors. Because each step is documented, an analyst who joins the chain later could rerun it without starting over, which is the practical value of treating data mining as a process rather than a one-time project.

Ethical Use of the Data

The public data used here are aggregates for whole ZIP codes and contain no information about individuals. When member data are added in the capstone, they will be summarized by ZIP code before analysis, so no individual member's address or attendance can be identified in any chart shared with the owners.

What the Owners Should Take Away

For the owners, the analysis changes the question. Instead of asking whether the next town over can support a studio, they can ask which of three specific ZIP codes in an underserved, affluent cluster offers the best site, a question the capstone will answer with rents, travel times and a demand estimate. The midterm's contribution is to narrow seventy possibilities to a handful worth studying in detail.

The capstone will add rents, travel times and competitors for the short list.

Limitations

Public data are averages for whole ZIP codes and lag by one to two years. Establishment counts include gyms, pools and racquet clubs that may not compete with boutique studios. The capstone project will add membership data from our own studios to test whether Cluster 2 residents resemble our current members.

Conclusion

Two public data sets, joined and clustered, turn a hunch about the next town into a defined set of candidate areas. The process of collecting, mining and presenting data here will feed the capstone's marketing plan.

References

Jain, A. K. (2010). Data clustering: 50 years beyond k-means. Pattern Recognition Letters, 31(8), 651-666. https://doi.org/10.1016/j.patrec.2009.09.011

Loshin, D. (2012). Business intelligence: The savvy manager's guide (2nd ed.). Morgan Kaufmann.

U.S. Census Bureau. (2024a). American Community Survey 5-year estimates. https://data.census.gov

U.S. Census Bureau. (2024b). County Business Patterns. https://www.census.gov/programs-surveys/cbp.html

Wedel, M., & Kannan, P. K. (2016). Marketing analytics for data-rich environments. Journal of Marketing, 80(6), 97-121. https://doi.org/10.1509/jm.15.0413