Data Analytics: Foundations to Practice · Data Cleaning and Preparation
Data Transformation and Structuring
Even clean, error-free data sometimes needs restructuring before it can actually answer the question being asked. This chapter covers the common transformations that reshape data for a specific analytical purpose.
Aggregation combines multiple individual records into a summarized value, such as converting individual daily transaction records into a single monthly total, which is often necessary because the specific business question being asked genuinely operates at a different level of detail than the raw, individual-record data was originally collected at.
Key Takeaways
- Aggregation combines individual records into a summarized value at the level of detail a specific business question actually operates at.
- Filtering narrows a dataset to relevant records; overly broad or narrow filter criteria can distort results without the distortion being immediately obvious.
- Data can be structured wide (categories as columns) or long (a category column paired with a value column); different tools expect one format or the other.
- Derived fields, calculated from existing fields and defined clearly upfront, improve consistency compared to recalculating the same logic informally each time.