What does data integration refer to?

Data integration refers to the process of combining and harmonizing data from different sources to provide a unified view. This is essential in data science, as data typically resides in various locations, formats, and structures. By integrating data, organizations can ensure that they have a comprehensive dataset that provides better insights and supports decision-making processes. This process often involves cleaning, transforming, and consolidating data to eliminate discrepancies and redundancies.

The other choices focus on aspects of data management that are not specifically about integration. Creating a backup of data pertains to data protection, which is crucial for preventing data loss but does not involve combining data. Storing data in a centralized repository is important for data accessibility and management but doesn’t address the process of combining different datasets. Performing data analysis on single datasets does not relate to integration, as it emphasizes working with one dataset instead of merging multiple sources. Therefore, the best definition of data integration is indeed the combination and harmonization of data from diverse sources.