Compare bottom-up to top-down modeling. What is the best approach to arrive at an effective data model?
The task to create an effective data model can be tackled from two opposite directions, which can be beneficial in certain circumstances. The top-down approach to database design is considered to be the most common (Kung, Kung, & Gardiner, 2013). The method suggests that the process should start with identifying populations, membership rules, and relationships between such populations (Kung et al., 2013).
The primary task of a database developer utilizing a top-down approach is to create an entity-relationship diagram that would that may not include all the attributes. In short, the primary concern of the top-down approach is to create a conceptual data model by identifying highly abstracted data objects, such as entity types and relationships between them.
The bottom-up method suggests starting the analysis from lower-level concepts, such as attributes and functional correlations. This approach aims at minimizing data redundancy and dependency from the beginning rather than normalizing after the initial planning is done. Both of the methods can be applied to database modeling; however, they serve different purposes. According to Kung et al. (2013), while no statistically significant differences in database design quality between the two methods were found, beginners are more successful in using a top-down approach, and experienced developers make fewer mistakes adopting the other one. Therefore, it may be concluded that there neither one of the approaches can be treated as best universally, as their success depends on the circumstances.
Kung, H. J., Kung, L., & Gardiner, A. (2013). Comparing top-down with bottom-up approaches: Teaching data modeling. Information Systems Education Journal 11(2), 14-24.