DATA 5003 Module 1 Business Question and Data Requirements Example

Why Do Chicago Restaurants Fail Inspection? Framing a Business Question and the Data to Answer It

Student Name

American College of Education

DATA5003: Data Analytics

Module 1 Assignment

Instructor Name

March 6, 2028

Introduction

I manage food safety and operations for a group of eleven fast-casual Mediterranean restaurants in Chicago, a composite company called Cedar and Salt in this course. In 2025, two of our kitchens failed routine city inspections, and one had to close for a day until a follow-up inspection passed. Each failure cost lost sales, overtime for deep cleaning and harm to our online ratings. The owners asked whether we are spending our food safety effort in the right places. This paper turns that concern into a business question that data can answer, identifies the data needed, plans the analysis to come in later modules and sets out its limits.

From Concern to Question

A worry such as are we safe enough cannot be analyzed. Provost and Fawcett (2013) describe data-driven decision making as basing decisions on the analysis of data rather than purely on intuition, and stress that useful analysis starts by framing a business problem precisely enough that data can inform it. The question for this course is therefore: which conditions and practices most strongly predict a failed City of Chicago restaurant inspection, and where should Cedar and Salt focus its coaching and spending to reduce the chance of failure? The question has a clear outcome, a pass or fail result, and a clear decision it serves, the allocation of a $90,000 annual food safety budget and managers' coaching time.

Why Data, and Why Now

Our current approach relies on managers' instincts and a generic checklist from a consultant. Brynjolfsson and McElheran (2016), using federal survey data, found that the share of American manufacturing plants using data-driven decision making rose sharply between 2005 and 2010, and that adopters tended to be more productive than similar plants that did not adopt. Restaurants are different from factories, but the logic carries over: if a public record shows which violations most often accompany failed inspections, we can direct effort there rather than spreading it evenly. Chicago publishes every food inspection result, which makes this question unusually answerable for a small company.

The Public Data

The City of Chicago's Department of Public Health publishes food inspection records on the city's open data portal (City of Chicago, 2026). Each record includes the establishment's name, license number, address and zip code, the facility type, a risk category assigned by the city, the inspection date and type, such as a routine canvass, a complaint or a license inspection, the result and the text of any violations, each with a number and description. A first query of the portal found 13,864 inspections of restaurants during 2025, of which 11,426 ended in a pass, a pass with conditions or a fail; the others ended with no entry, a business out of operation or a kitchen not ready.

Our Own Data

The city data describe outcomes and violations, but not the practices behind them. Cedar and Salt's own records add four sources: weekly self-inspections by kitchen managers, scored on a 40-item checklist since 2023; pest control service reports from our contractor; staff training records showing food handler and allergen training dates; and maintenance tickets for equipment such as refrigerators, dish machines and hand sinks. Linking these to our own city inspection results will show whether, for example, kitchens with overdue maintenance tickets or lapsed training are more likely to receive the violations that predict failure. All four sources are kept in different systems today, which is the first practical obstacle Module 2 must solve.

Data Quality Concerns

Both sources have quality problems that the analysis must handle. The city's violation field is free text that combines a numbered category with inspectors' comments, so violation numbers must be extracted carefully, and the city changed its violation categories in 2018, which makes older records hard to compare. Some establishments appear under slightly different names or addresses, though the license number provides a more reliable key. Facility types are recorded inconsistently, so the analysis will limit itself to records coded as restaurants. Internal self-inspections were completed less consistently at two kitchens during staff shortages in 2024, which means missing weeks rather than poor scores, and the analysis will treat those weeks as missing rather than as zero.

What a Useful Answer Looks Like

Before analyzing anything, it helps to define the answer that would change a decision. A useful result would rank the handful of violations most strongly associated with failing, show how large the difference in failure rates is for each, and link at least some of them to practices we control, such as pest service frequency, hand sink maintenance or training. If, for example, inspections that find evidence of pests fail far more often than those that do not, and our own pest reports show gaps at certain kitchens, the budget decision becomes concrete. A result that simply lists every violation in order of frequency would not help, because frequent problems are not always the ones that cause failure.

Analysis Plan

The analysis will proceed in the order of this course. Module 2 will describe the analytics system: where each data source lives, how it is collected and how it will be joined. Module 3 will apply descriptive statistics and visualization to the 2025 city data, showing failure rates by inspection type, risk category and month, and the most common violations. Module 4 will look for relationships, comparing failure rates for inspections with and without each common violation and testing whether differences are larger than chance would explain. Module 5 will turn the findings into a recommendation for spending and coaching.

Limits of the Question

Three limits should be stated at the outset. First, inspection results are not the same as food safety. Jones et al. (2004), comparing restaurant inspection scores in Tennessee with foodborne disease outbreaks, found that restaurants with outbreaks did not have markedly worse inspection scores, suggesting that inspections capture only part of the risk. Our goal is to avoid failed inspections and the risks they flag, not to claim that passing guarantees safety. Second, the city data reflect when and where inspectors visit, which depends partly on complaints. Third, our eleven kitchens are a small sample, so internal comparisons will be suggestive rather than conclusive.

Ethical Considerations

The city data are public, but they concern real businesses, including competitors, so the analysis will report patterns across all restaurants and will not single out other establishments by name. Internal data include employee training records, which will be analyzed by kitchen rather than by individual so that the work does not become a tool for disciplining staff. Results will be shared with kitchen managers before they are shared with the owners, so managers can explain context the numbers miss.

Conclusion

The owners' concern about food safety has become a specific question: which conditions and practices predict a failed Chicago inspection, and where should the group focus its budget and coaching? Public inspection records for more than 11,000 restaurant inspections in 2025, combined with the group's self-inspections, pest, training and maintenance records, can answer it. The plan for the remaining modules moves from describing the data to finding relationships and making a recommendation, with its limits stated from the start.

References

Brynjolfsson, E., & McElheran, K. (2016). The rapid adoption of data-driven decision-making. American Economic Review, 106(5), 133-139. https://doi.org/10.1257/aer.p20161016

City of Chicago. (2026). Food inspections [Data set]. Chicago Data Portal. https://data.cityofchicago.org/Health-Human-Services/Food-Inspections/4ijn-s7e5

Jones, T. F., Pavlin, B. I., LaFleur, B. J., Ingram, L. A., & Schaffner, W. (2004). Restaurant inspection scores and foodborne disease. Emerging Infectious Diseases, 10(4), 688-692. https://doi.org/10.3201/eid1004.030343

Provost, F., & Fawcett, T. (2013). Data science and its relationship to big data and data-driven decision making. Big Data, 1(1), 51-59. https://doi.org/10.1089/big.2013.1508