Which transformation is appropriate to create derived fields using data from both current and previous rows?

The computeRelative transformation is specifically designed for creating derived fields using data from both current and previous rows. This transformation allows for the reference of values from prior rows, enabling users to calculate metrics or derive fields that depend on historical data within the dataset.

For example, it can be utilized to calculate running totals, moving averages, or any metric that requires an understanding of previous records in relation to the current row. This functionality effectively enables advanced analyses and dynamic calculations that are crucial for gaining insights from time-series data or sequential records.

On the other hand, other transformations serve different purposes. The digest transformation primarily focuses on summarizing or aggregating data, which does not incorporate the relational aspect of prior and current row comparisons. The augment transformation adds additional context or information from linked datasets rather than manipulating row-based data within the same table. Lastly, the append transformation is used to combine two datasets into one but does not create derived fields in relation to the sequence of data. Thus, computeRelative stands out for its ability to directly utilize both current and previous rows for deriving new insights.