In most datasets, each row belongs to a different element such as a different sale or a different customer. However, there are datasets where a single row doesn't completely describe one element. Take, for example, the file from Chapter 8, Manipulating Data by Coding, containing information about houses. Every house was described through several rows. A single row gave incomplete information about the house. The ideal situation would be one in which all the attributes for the house were in a single row. With PDI, you can convert the data to this alternative format.
The Row denormaliser
step converts the incoming dataset to a new dataset by moving information from rows to columns according to the values of a key field.
To understand how the Row denormaliser
works, let's introduce an example. We will work with a file containing a list of French movies of all times. This is how it looks:
... Caché Year...