During my analysis of U.S. military troop deployments in Europe from 2006 to 2015, I focused on exploring patterns across countries and branches using SQL. While the dataset appeared clean, I implemented a simple validation check: the total number of troops in each row should equal the sum of all branches—army, navy, air force, and marines.

1. Discovery of a Minor Anomaly

   
   Most rows passed the validation, but the United Kingdom in 2014 revealed a discrepancy of just one troop. Although trivial in size, this small mismatch piqued my curiosity and highlighted how even curated datasets can harbor subtle errors.

   
   

2. Investigating the Cause

   
   I verified all data types, ensured no nulls were present, and examined the values for inconsistencies. The likely cause was either human oversight during manual entry or independent adjustment of the “total troops” column. This investigation reinforced the importance of scrutinizing data, even when it appears reliable.

   
   

3. Adjusting Validation Rules

   
   To handle minor discrepancies gracefully, I modified the validation logic to allow a margin of one troop. This approach ensured that harmless inconsistencies were ignored, while still flagging significant errors elsewhere in the dataset.

   
   

4. Key Takeaways

   
   The project emphasized a critical principle: real-world data is rarely perfect. Totals may not match their parts, and assumptions can easily break. Small anomalies, like a single missing soldier, serve as reminders to consistently validate, challenge, and question datasets before drawing conclusions.

   
   

5. Additional Resources

   
   The full project, including all queries, cleaning logic, and further insights, is available on GitHub. Engaging with such projects strengthens practical data analysis skills and prepares for more complex data-driven tasks in real-world scenarios.